As well as offering social and health benefits, water is also a key ingredient for sectors such as agriculture and energy production. Discontinuous supply challenges managers to rely on predicting water flow well in advance, to avoid service interruption or overflow. To reduce the impact from extreme events and help customers adapt to new climatic conditions, the EU-supported Waterjade project developed a system which can predict water flow from all river outlets worldwide. Waterjade’s new prediction algorithm blends physical models with artificial intelligence to track the whole water cycle, from snow in the mountains to rivers in the valley, with higher accuracy.
The prediction algorithm
Currently, water prediction typically relies on simplified empirical models based on historical data which cannot capture fast-moving climate trends and hydrological cycle changes. This can result in error-prone forecasts which can have serious consequences. One such example is the cost to businesses in 2018, when the Rhine experienced one of its longest dry spells on record. These costs, and others like them, could have been mitigated with better prediction. Waterjade’s prediction tool combines water plant specifications with historical data and meteorological data to inform its algorithm, which works with physical models, to predict water inflow in rivers and reservoirs. The system uses data from a range of Open Data sources, such as in situ meteorological stations, satellite images from the Copernicus project and numerical weather predictions. Customers logged in to the Waterjade platform can find water forecasts for their plants, for forthcoming days and months. A decision support system is also available, to help them manage reservoir levels and identify possible forthcoming problems, such as droughts and floods. Users can take advantage of optimisation tools which visualise available water, charting it against demand predictions, based on different scenarios determined by the user. “The uniqueness of our solution lies in its flexibility for a range of local contexts, such as reservoir configurations, along with its accuracy regarding the impact of a range of changing climate conditions,” says Matteo Dall’Amico, project manager. So far, the team have tested their solution in two different climatic conditions: the Alps, where the dominant inflow to rivers comes from snow, and southern Spain, characterised by a warm and arid climate. “In both cases, Waterjade improved prediction accuracy, reducing errors by 50 %, compared with state-of-the-art techniques, in terms of snow monitoring and river inflow predictions,” says Dall’Amico.
Tackling water scarcity
According to the European Commission, about 10 % of the total EU area is subject to water scarcity. The direct economic impact of drought events was calculated at a minimum of EUR 100 billion over the 30 years prior to 2007. “By anticipating crises, Waterjade offers a tool that can help water managers, particularly of hydropower facilities, mitigate water shortages, rather than simply respond to them,” explains Dall’Amico. With a patent application soon underway, Waterjade is currently subscription-based, but the team aim to introduce a Software as a Service option. The team are now working to augment the system’s seasonal forecasting and snow monitoring capability by improving satellite image analysis. They are pursuing advanced downscaling techniques for high-resolution coverage of local conditions. Waterjade is also developing a platform dashboard for customer interaction. To bring the system to market, which they anticipate includes around 20 000 European utilities, the team are looking for potential partners.
Waterjade, drought, reservoir, global warming, weather, hydrological, forecast, snow, river, meteorological, floods