From October 7–11, 2024, Romee van Dam from Deltares traveled to UNIVERSITA DEGLI STUDI DI ROMA LA SAPIENZA in Rome to collaborate with Marco, Sathish, and Valeria on the development of (AI) groundwater quality modelling as part of the NINFA project. The focus of the visit was Case Study 1 (Baakse Beek) and Case Study 2 (Los Alcázares), with the aim of integrating traditional groundwater transport modelling (WP2) with advanced AI approaches (WP4) to enhance predictive capabilities.
The visit facilitated the transfer of information from WP2, led by the Rome team, to WP4, led by Deltares. WP2 involves building a detailed transport model for groundwater quality using MODFLOW and solute transport software to simulate conditions in the Baakse Beek catchment. In WP4, Deltares is developing a simplified version of this model using MODFLOW6 and MT3D to quickly generate 10,000 simulation results. These datasets include variables such as precipitation and evapotranspiration under climate change scenarios, which will be used to train an AI model capable of producing results in mere minutes, compared to the hours required by traditional methods.
During the visit, the team worked to finalize the simplified transport model, ensuring its consistency with the full transport model. Romee and her colleagues ran initial tests of the simplified model, including simulations of a hypothetical contaminant’s transport within the Baakse Beek catchment. The results demonstrated promising progress toward creating a robust dataset for AI training.
The discussions in Rome also focused on harmonizing methods to align the simplified and full transport models, ensuring comparable results. The week proved both productive and inspiring, marking a significant milestone in the AI modelling of groundwater quality.
Romee expressed her gratitude for the hospitality and collaboration in Rome, emphasizing the importance of teamwork in driving innovation. The visit further strengthened connections between project partners and laid the groundwork for future advancements in the NINFA project.