Objectives
The project aims to develop and validate a prototype quantitative agro-climate risk indicator for the maize, soft wheat, coffee and sunflower seed markets, reaching Technology Readiness Level 7.
The project is structured into five work packages:
- collection and post-processing of satellite and meteorological reanalysis data;
- development of a weather forecasting framework with a 12–18-month horizon;
- identification of clusters and forecasting scenarios;
- development of agro-climate indices;
- development and validation of the prototype indicator using data from 2022 to 2025.
The project is expected to last 11 months.
Initial Challenge
Companies operating in the agri-food sector are highly exposed to volatility in agricultural commodity prices. Existing forecasting tools are largely based on qualitative methodologies, including surveys and on-site assessments. They are generally updated no more than once a month and are not directly linked to weather forecasts.
There is also a lack of forecasting capabilities that can incorporate global weather outlooks over a sufficiently long time horizon. As a result, companies often lack timely and effective risk-mitigation strategies.
Solution
To address this challenge, the project develops a prototype agro-climate risk indicator that integrates AI-generated seasonal weather forecasts with the fundamental production drivers of agricultural commodities. The indicator estimates, up to 12–18 months in advance, the probability that critical weather and climate thresholds with a potential impact on prices will be exceeded.
The solution:
- uses satellite datasets, meteorological reanalysis data and global agro-climate indicators;
- applies clustering algorithms to identify reference scenarios;
- employs a hybrid deep learning architecture combining a recurrent module for trends and seasonality with a decision-tree-based module for non-linear relationships.
The indicator is validated against real-world data from 2022 to 2025 for the Italian market.
Benefits
CropCast delivers measurable benefits for all stakeholders exposed to agro-climate risk:
- timely, quantitative and potentially real-time forecasts and indicators, offering greater accuracy than existing tools;
- the ability to anticipate the impact of weather conditions on prices, with updates available as frequently as daily;
- improved purchasing and sales decisions across the entire agri-food supply chain;
- support for more accurate hedging strategies to manage price volatility;
- new opportunities for financial, banking and insurance operators, including credit ratings and parametric insurance products;
- increased resilience and competitiveness across the agri-food sector.
IFAB’s Role
IFAB played a dual role in the project: as a funder, selecting CropCast within its research and innovation investment programmes for its scientific and practical value, and as a technical partner actively contributing to the development of the artificial intelligence methodologies underlying the agro-climate risk indicator.
From a technical perspective, IFAB’s contribution covered the project’s entire methodological framework. The data foundation was built by harmonising satellite datasets, reanalysis data and global agro-climate indicators, and integrating them with seasonal weather forecasts generated by artificial intelligence models over a 12–18-month horizon. Clustering algorithms were then applied to identify reference scenarios and interpret the variability of agro-climate conditions across the maize, wheat, coffee and sunflower markets. At the core of the system is a hybrid deep learning architecture that combines a recurrent module for capturing trends and seasonality with a decision-tree-based module for modelling non-linear relationships.
Participation in CropCast enabled Fondazione IFAB to strengthen its expertise at the intersection of climate forecasting, time-series modelling and financial risk analysis in commodity markets.
For further information, please contact: projects@ifabfoundation.org
