Speaker
Description
The construction of future large-scale research infrastructures (e.g., FCC, Einstein Telescope) will generate significant volumes of excavated materials that require valorization. One sustainable approach to managing these materials is their use in reconstituting functional soils. This presentation outlines the planned activities of Work Package 4 (WP4) of the IRIS project (Intelligent Research Infrastructure Sustainability), which focuses on evaluating soil-plant functions and ecosystem services in reconstituted soils using AI-based surrogate models. To achieve this, data collected from pilot laboratories, such as the OpenSkyLab, will be leveraged. The OpenSkyLab is a research infrastructure designed to test the functionality of various soil-crop reconstitution treatments. Key soil and crop functions, as well as ecosystem services, will be monitored within this facility.
The WP4 activities will follow a structured approach: i) Desk Research: Identify operational procedures for soil reconstitution and select appropriate process-based soil-crop models to study soil-plant functions and ecosystem services; ii) Model Selection and Calibration: Choose a process-based model and calibrate it for the different treatments tested in the OpenSkyLab; iii) AI-Driven Scenario Analysis: Use the calibrated model to conduct scenario analyses, enabling the evaluation of soil functions and services beyond the OpenSkyLab’s testing horizon; and iv) Surrogate Model Development: Design a simplified AI-based surrogate model that can be integrated into operational soil reconstitution procedures. Model validation will be performed using soil functional quality indicators derived from benchmark datasets covering natural and anthropogenic soils across France and Switzerland, ensuring robustness across contrasting soil systems. The validation framework will be further refined based on the availability and selection of reference indicators from experimental measurements and literature sources.
The knowledge and expertise generated through this project will be used to train operational managers of large research infrastructure plants and will be disseminated to various end-users to maximize impact.