AI for dynamic water resource management – AIqua

Climate change is profoundly altering the way in which hydrosystems function, leading to a gradual scarcity of water resources in summer and, in some areas, an increase in winter. This context calls into question current models for managing these resources, which are designed for situations of abundance and based on historical benchmarks that are becoming less and less relevant. What's more, they focus on a reactive approach, which limits the ability to respond to water shortages or surpluses. Decisions are often taken too late, exacerbating the impact of droughts or floods. In addition, this reactive approach means that the volumes that can be abstracted are limited to what is available in the driest years, unnecessarily restricting access to the resource and making users more vulnerable, without offering sustainable solutions to future climate variations.
Against a backdrop of increasing pressure on resources, water stakeholders are looking for new management models capable of responding more effectively to hydro-climatic fluctuations. The AIqua LabCom is positioning itself within this context to propose dynamic management methods based on anticipation and artificial intelligence. Dynamic management involves adapting management strategies in real time, optimising the use of resources according to observed and forecast conditions, to ensure greater resilience in the face of climatic hazards and offer greater flexibility to meet the varied needs of users (agriculture, industry, environment).
The aim of the AIqua LabCom is to design the forecasting algorithms needed to calculate the decision indicators required for this dynamic management. The approach is fully in line with Sustainable Development Goal 6 ‘clean water and sanitation’. The purpose of these algorithms is to model and forecast in real time the complex interactions between climate, hydrology and human activities, but also to help decision-makers interpret the modelling results and calibrate their decisions. They are based on innovative digital technologies, in particular artificial intelligence and big data, applied to hydrological modelling, as well as theories of decision-making under uncertainty.
AIqua brings together the Gustave Eiffel University's GERS laboratory and Aquasys, a publisher of digital tools for water management. The GERS laboratory is contributing its expertise in hydrology and algorithms, while Aquasys is contributing its knowledge of decision-making needs and its know-how as a publisher of digital services for water management. Together, the two partners aim to make dynamic management more accessible by harnessing the power of AI and its capacity for automation.
In scientific terms, AIqua will focus on the design of hydrological models based on multi-scale and multimodal data, enabling (1) better hydrological modelling on several geographical scales, (2) the development of models for different types of data (surface water, groundwater, etc.) for forecasting horizons ranging from one day to one season, (3) the characterisation of modelling uncertainties incorporating anthropogenic pressures on the resource, enabling the development of decision-support indicators.
The aim will be to identify the best modelling strategies, making the most of AI technologies and taking into account best practice in design, development and software integration, to facilitate the industrialisation of digital solutions. The final aspect of the LabCom project will be to convert the data generated by these AI models (forecasts and associated uncertainties) into decision-making indicators for stakeholders with different profiles and interests.