The development of an intelligent sensor system for dykes promises to dramatically reduce maintenance and repair costs by detecting signs of damage at an early stage.

The system, developed by German engineering giant Siemens, consists of a network of sensors situated both above and below the surface of the water. That network collects a range of data on environmental conditions, including temperature, pressure and humidity within the dyke, as well as the temperature and depth of canal water.

This information is dispatched to a central control hub via GPRS, where it is processed and compared to historical data by means of  an intelligent neural network system in order to best discern signs of damage.


The neural network system is highly adept at recognising any atypical deviation from historical data which could serve as an early indicator of damage.

Examples include discrepancies of temperature within the dyke itself, which could indicate that the levee is on the verge of bursting due to an ingress of warmer water, or the water table remaining at an unusual height compared to historical data for prevailing levels of precipitation.

By recognising such atypical deviations, and distinguishing them from regular deviations that are not symptomatic of damage, the neural network can discern problems weeks or even months before their actual onset.

This form of automated monitoring could reduce maintenance costs for dykes by as much as 10 to 20 per cent – an impressive benefit given the increasing importance such infrastructure as global warming lifts sea levels and flooding becomes more frequent.

Siemens’ intelligent monitoring system has already passed muster after field testing in the Netherlands, a nation renowned for the sophistication of water and dyke engineering.

Waternet Amsterdam – which is responsible for 1,000 kilometres of dykes safeguarding a population of one million occupying 700 square kilometres of land – was the pilot customer for the trial, installing a network of sensors along five kilometres of a dyke in Amsterdam.

The sensors were installed in the dyke at distances of approximately 100 metres from each other both above and below the waterline.

In order to extend the operating life of the sensors data is issued at intervals of an hour, although under hazardous conditions this rate can be increased to reports by the minute.