Forecasting drought risks: artificial intelligence comes to the rescue of local authorities

France is now experiencing more and more water stress every year. As a result, local authorities must learn to manage their water resources in new ways. The professional solutions available to them include the powerful forecasting tool produced by Saur Group company imaGeau.

Right around the world, the availability and quality of water are under threat. On the basis of these criteria, the World Resources Institute ranks France as a ‘medium risk' country, alongside Germany, China and Peru. But in recent years, our country has nevertheless experienced a series of serious drought events, and as global warming intensifies, meteorologists are predicting longer and hotter summers in future years. The inevitable result will be even higher levels of water stress.

Local authorities have, for some time, recognized the essential need to understand how to limit the risks related to water resources through conservation, protection and sustainable use. Some experts - including those at imaGeau, the company set up by Saur Group in 2009 - are now going further still. Having made the ongoing protection of water resources their specialty, they are now also working on forecasting drought risks to help local authorities in their planning and management.

Sustainable long-term management of local authority water resources

Take the example of water tables - although the same principle applies equally to rivers - whose levels rise in winter, and fall in summer. Charted on a graph, the trend is a waveform. In France, overexploiting groundwater resources is never an option, since all the authorities involved are extremely vigilant to ensure that it cannot happen. When a water table gets too close to its historically low level, drought decrees are issued to limit abstraction.

So local authorities use groundwater levels to protect them and the amount of water abstracted from them. But being able to forecast future trends in groundwater levels gives operators a clear advantage in terms of making the right decisions for sustainable, long-term smart management of their water resources.

Leveraging the power of artificial intelligence for better water resource management

These are the reasons why imaGeau has built a software solution that can not only keep local authorities up to date with daily fluctuations in groundwater levels, but also forecast future levels. Based on machine learning, it is able to concentrate data around the many variables that influence groundwater levels, including rainfall, interactions with rivers and soils, temperatures, irrigation and soil water content; the interactions between which are impossible for human forecasters to predict accurately.

All the data generated provides input for artificial intelligence algorithms similar to those that have already been in use for several years in financial analysis and the digital world, where they are extensively applied by tech giants like Google. AI systems ‘learn’ the interactions between all these datasets, and then try to reproduce the observed signal. They then selects the function that provides the closest match between the forecast signal and the observed signal. This function is then used to forecast drought risks and the future trend in groundwater levels.

The solution is currently being rolled out with a number of water agencies, including the Syndicat d'exploitation des eaux des Landes, the Syndicat d'exploitation des eaux de la Manche and the Montauban urban authority.

These local authorities now have direct access to scenarios forecasting water resource trends for up to 90 days. The forecasting window for rivers is 20 days, since they are more immediately responsive to rainfall levels. As a result, they are able to forecast drought risks and put in place action plans designed to avoid water supply crises. The effectiveness of the decisions made can then be assessed daily, and the model adjusted accordingly.

To find out more, contact us: