ESA title

Floodly AI

  • ACTIVITYKick-start Activity
  • STATUSCompleted
  • THEMATIC AREAInfrastructure & Smart Cities, Environment, Wildlife and Natural Resources

Objectives of the service

The proposed service will:

  • Survey water utilities entire water pipe networks and locate all of their pipe leaks. Quantify and rank each leak.
  • Find underground water sources using satellite data and AI.  Empower water charities/non-profits with the ability to survey areas of interest quickly and accurately.

Image credit: DeepWaters AI/Abdul Daud

Users and their needs

In the world's poorest countries over 1 billion people do not have access to clean drinking water, in the richest countries, ageing pipes leak billions of litres of water.

Identifying new sources of underground water is important for governments and water utilities in developing countries.  This is also of importance for water related charities, non-profits and NGO’s working in those countries.

Locating new sources of underground water is also of importance to governments and water utilities in developed countries.  Many developed countries such as the USA and the UK also face serious water shortages and droughts.

Given the issues of water scarcity in many countries, water pipe leaks also represent a serious issue.  In countries such as the USA and UK, corroding pipe infrastructure can be as much as 100 years old, these pipes leak billions of litres of water every day.  

Service/ system concept

  • Combine multi-spectral and synthetic aperture radar satellite data with neural networks.
  • For underground water location surveys entire patches of terrain are analysed.
  • For water pipe leaks, specific areas of terrain related to pipe infrastructure network locations are analysed.

Space Added Value

The solution combines neural networks with ESA Sentinel 1 & 2 satellite data.  The neural networks are trained to identify the spectral signature of water and backscatter propagation signature of increased surface moisture.


Current Status

Image credit :  SpaceWater.AI, DeepWaters.AI, Massoud Maqbool

  • A map of the Earth’s sub-surface water was created at a spatial resolution of 10msq.  This required surveying over 1.5 trillion, 10msq ‘satellite tiles’. Each tile stores 50 satellite and bespoke data parameters; over 300Tb of data.
  • Launched a free underground water mapping service SpaceWater.AI, with support from Esri, Nvidia and Amazon Web Services. Pilot users include UNHCR and WaterAid.  
  • Underground water location identification, maximum peak accuracy is up to 98% depending on geographic and environmental conditions.
  • Peak pipe leak location identification accuracy is at prototype validation stage. 

Prime Contractor(s)

Status Date

Updated: 15 January 2021