Water leaks discovered with drones and camera sensing technology
Image credit: Dreamstime
Rapid surveys of water networks to detect leaks can now be achieved thanks to remote sensing cameras strapped to the underside of small aircraft.
A recent report by the European Commission found that about half of the EU countries lose more than 20 per cent of their clean drinking water before it reaches consumers’ taps, and in some countries the proportion is as high as 60 per cent.
Often, old piping systems and poor water pressure management are to blame and many leaks go undetected because they are not visible on the surface.
Rapid inspection of pipes is now being made possible by taking pictures with remote sensing cameras, mounted on a small plane or an unmanned aerial vehicle (UAV).
The images recorded in different spectral bands, from the visible to the short wave infrared and in the thermal infrared band, are then interpreted in labs.
“The main information is obtained in the thermal infrared band. Leaks will be detected through changes in the near environment, essentially through the increased soil moisture and maybe through changes in vegetation,” said Jean-Claude Krapez, senior research scientist at the French aerospace lab ONERA.
He added that one camera actually monitors the thermal radiation emitted by the soil.
“Compared to dry areas, wet areas are subject of latent heat exchange with the atmosphere, resulting in a cooling of the surface. This is seen in our images, as darker areas, which are correlated to lower temperature,” Krapez said.
Nevertheless, the thermal radiation level depends not only on temperature, but also on the earth’s surface material, either soil, or rocks, and on the vegetation present.
Several tests have so far been undertaken in the Provence region in the south of France, using on-board cameras on a small aircraft. Further surveys are envisaged with smaller cameras installed on an unmanned aerial vehicle.
“Both camera systems will work at approximately the same resolution, they will collect the optical signals in nearly the same bands, so we expect the images to be quite similar. With our aircraft we can inspect more linear systems, such as long pipes running for several kilometres, while with the UAV we will inspect more complex networks with many branches and short lines,” he said.
Tens of kilometres of rural land can be surveyed by aircraft equipped with on-board cameras. “It can be different types of water transportation means, either underground, pressurised pipelines or open air canals, drinking water pipelines or water transportation for irrigation,” explained Krapez.
However, some challenges remain when it comes to images interpretation, due to the high heterogeneity of the surface: bare soils, low grass, meadows, crops, bushes, high trees.
“There are many other causes of temperature variations than moisture changes. Therefore, there are a lot of ambiguities,” said Krapez. To solve them, at least partially, the scientists unravel the apparent temperature readings with the help of a vegetation index and analyse the pixels from the remote sensing images in a multidimensional space.
Once the data has been processed, they obtain a ‘water index map’. But even at this point, images are difficult to interpret, as there are still spurious contrasts and shadow effects.
A high water index area could be a real leak or even a puddle, or, in other instances, tree shadows could be mistaken for wet zones. “It will be necessary to interact with people from the water distribution companies to avoid false signals,” Krapez added.
A team recently demonstrated an AI controller that helps to reduce water leakage in pipe networks by optimising the pump rate based on several variables such as local temperature, time of year and whether it is vacation season.