HybridCrop pioneers a new generation of hybrid crop models that combine the interpretability and transparency of mechanistic models with the predictive power and computational efficiency of machine learning to improve agricultural risk assessments under climate extremes.

While mechanistic crop models provide biophysically consistent representations of plant growth processes and their interactions, they exhibit structural limitations, including high computational costs, parameterization burdens, and biases—particularly under extreme conditions such as droughts and heatwaves. Machine-learning models, in contrast, can achieve high predictive accuracy but are prone to overfitting and often lack process understanding, interpretability, and transferability under future climate conditions.

HybridCrop addresses these challenges through a genuinely new modelling paradigm in which specific biophysical processes are represented by physics-informed machine-learning components that retain mechanistic interpretability while overcoming structural constraints of existing crop models. Instead of relying on black-box yield prediction, machine-learning modules are trained to represent fundamental crop processes such as leaf area development, phenology, evapotranspiration, and soil moisture dynamics. These modules are integrated into well-established crop models (EPIC and LPJmL), creating process-oriented hybrid systems that preserve biophysical consistency while improving predictive skill and computational efficiency. In doing so, HybridCrop aims to redefine how process understanding can be combined with data-driven methods. To ensure transparent and reproducible model evaluation, the project will develop and release open-access benchmark datasets based on Earth observation and in-situ data, enabling systematic comparisons between hybrid and conventional crop models.

A unique feature of HybridCrop is the use of Single Model Initial-condition Large Ensembles (SMILEs)—large ensembles of climate simulations that enable the systematic analysis of rare and compound climate extremes with unprecedented statistical robustness. The new hybrid modelling frameworks will be driven with SMILEs to generate new insights into agricultural risks in a future climate characterized by increasing variability and extremes.

Together, these innovations will establish a new methodological framework for crop modelling and climate impact assessment, enabling more reliable, transparent, and computationally efficient analyses of agricultural risks under a changing climate. By combining process-oriented machine learning with large climate ensembles, HybridCrop will set a new standard for assessing agricultural risks and adaptation options in a future shaped by increasing climate variability and extremes.

BESTMAP is an international research project run by a consortium of 13 partners across 7 countries and funded through the European Union's Horizon 2020 Research and Innovation Programme.

The main goal of BESTMAP is to improve and contribute to the existing tools used in agricultural policy impact assessment and to develop a new model-based approach to simulate the environmental impact of changes in the CAP, in order to promote a sustainable future for the EU’s agricultural sector.

The Research Group “Global and Regional Land-Use Change” contributes to the project by running scenarios on EU’s biofuel policies and international climate policies. Further, it will elaborate a concept to link economic modelling with individual-farm Agent-Based Models (ABM) aiming to improve the modelling of complex farmers' decision making in economic models.

For more information please visit the project’s website.

ReFuel.ch (Renewable Fuels and Chemicals for Switzerland)

To meet Swiss net-zero targets, accelerated market development of sustainable fuels and platform chemicals (SFPC) is needed. The ReFuel.ch consortium, funded by the SWEET (SWiss Energy research for the Energy Transition) program of the Swiss Federal Office of Energy (SFOE), brings together nine Swiss universities, research institutes, and industrial partners. This interdisciplinary effort will involve contributions from social sciences, natural sciences, and engineering, as well as ongoing dialogues with policymakers, market actors, and end-users. The consortium's goal is to develop robust consumption and supply paths for SFPC that align with long-term climate policy objectives. The ReFuel.ch project is built around seven working packages, and the “Global and Regional Land-use Change” research group of the University of Basel is leading the international social, economic, and policy assessment.

Indeed, a national strategy for a sustainable and climate-friendly energy system is closely linked to social, economic, and policy factors on the international level, particularly if it involves significant energy trade, as may be the case for Switzerland. Since most SFPCs are expected to be imported, the implications on international markets, social impacts in exporting countries, and (in)direct environmental impacts are uncertain and dependent on the technology used. Therefore, the goal is to assess different pathways while carefully considering these interdependencies in the context of the international policy environment.

• Identify and quantify impacts on selected exporting countries

  Focusing on two potential exporting countries, Computable General Equilibrium (CGE) models are used to analyze the socio-economic and environmental impacts of supplying SFPC or their feedstocks to Switzerland. Spain, identified as a promising provider of renewable electricity and storage products like methanol, hydrogen, and sustainable aviation fuels (SAF), will be the first case study, with policy scenarios modeled based on findings of other working packages. A similar approach will be applied to a second SFPC exporting country, assessing how increased Swiss demand could impact its economy, society, and use of natural resources.

• Identify and quantify impacts on international agricultural markets

  To meet the global demand for SFPC, large-scale deployments of renewable energy production require a considerable amount of electricity and natural resources including land and water. This increasing need for natural resources could impact the international commodity market, influencing the global economic structure. These complex interactions among economic agents will be assessed through CGE models in order to capture the impact of technology diffusion and to simulate different policy interventions.

For more information about the project visit Renewable Fuels & Chemicals for Switzerland | SWEET reFUel.ch

Contacts for more information: Ruth Delzeit, Robin Argueyrolles, Chun-Yu Chen, Gianna Angermayr