Weather radar data and AI: an open data infrastructure for nowcasting in Italy

In its first year of operation, IT4LIA – AI Factory developed a new data infrastructure for monitoring and forecasting rainfall in Italy, tackling a complex challenge: overcoming the historical fragmentation of national radar data to make it accessible and ready for use.

IT-DPC-SRI: Italy’s first radar “Data Cube”

The project, coordinated by Fondazione Bruno Kessler, led to the creation of IT-DPC-SRI, the first public, standardized archive of precipitation radar observations covering the entire national territory. The project’s numbers highlight an unprecedented resource:

• Historical coverage: 16 years of data to date, starting in 2010.
• High resolution: 1 km spatial resolution.
• Continuous updates: new data integrated every 5 minutes.
• Volume: over one million measurements already available.

The initiative also fills a strategic gap in meteorological radar monitoring in Europe. Italy, in fact, has never joined OPERA, the European-wide exchange program for radar observations, making sure that the continental composite radar lacks coverage on our national territory. Thanks to this new dataset, Italy can finally be integrated into European-scale analyses, bringing the country in line with the most advanced international standards.

Analysis-Ready data: from information to action

The innovation lies not only in the quantity and uniqueness of the data released, but above all in the way they are made available to users.  The  data were collected, preprocessed, and converted into an analysis-ready format designed for use on cloud infrastructures. Information that was initially scattered across different formats and difficult to combine has been standardized and organized into a single data cube that is easy to access and use, including by artificial intelligence models. This step significantly reduces technical complexity, allowing companies, public bodies, and researchers to work directly with the data without having to build complex data preparation procedures.

The dataset is accessible through several complementary channels:

• Static version (Zenodo): designed to support the reproducibility of scientific studies and analyses; covers the period from 2010 to 2025.
• Static version with interactive access (European Weather Cloud): enables interactive and segmented access, eliminating the need to download the entire archive; covers the years from 2010 to 2025 and can be accessed through the Python mlcast-dataset package, developed and maintained by the MLCast scientific community of the EUMETNET E-AI Optional Programme.
• Continuously updated version (ArcoDataHub): provides “live data” updated in real time every five minutes to support immediate operational applications; covers the period from 2010 to the present and enables interactive, partitioned access.

A strategic asset for the market: value for companies

The availability of such a rich historical archive, combined with a real-time data stream at kilometer-scale resolution, represents a major competitive and operational advantage for the industrial and service sectors.  To date, the difficulty of accessing homogeneous radar data has slowed the development of proprietary risk-management and optimization solutions. IT-DPC-SRI unlocks this potential across several key sectors, for example:

• Smart agriculture: cooperatives and farms will be able to use radar data for precision irrigation, monitoring rainfall accumulation in fields to prevent water stress or damage caused by extreme events.
• Water network management: water utilities will be able to integrate these data into hydrological models to optimize water collection, prevent waste, balance storage reservoirs, and efficiently monitor urban drainage systems.
• Infomobility and routing: transport companies, couriers, and logistics companies will be able to use real-time radar maps to reroute vehicles and avoid intense storms, flooding, or poor visibility, optimizing fleet management.
• Insurance sector: companies will be able to rely on historical, standardized datasets to quickly and objectively quantify damage from extreme rainfall, accelerating appraisals and claim settlement at the local level.
• Renewable energy: hydropower producers will be able to use rainfall data to predict water inflows into reservoirs and schedule production in advance.

Integrating these open data into business workflows no longer requires lengthy and expensive data-cleaning processes. Because IT-DPC-SRI is natively Analysis-Ready, it drastically reduces time to market for companies seeking to develop new digital services.

This standardized database is also an ideal dataset for developing artificial intelligence models for very short-term forecasting (nowcasting). Companies can train predictive algorithms and apply them to continuously updated data in real time, enabling automatic alerts in their systems and protecting production sites, construction sites, and warehouses before torrential rainfall reaches the area.

Coming soon, IT4LIA will release its own open-source AI-based model for predictive nowcasting, named IRENE. This innovative technological resource and its business applications will be the focus of an in-depth analysis in our next article.

A partnership of excellence enabled by IT4LIA

The work is the result of a broad collaboration among leading institutions and research centers: Fondazione Bruno Kessler, the Department of Civil Protection, Agenzia ItaliaMeteo, ARPAE Emilia-Romagna, and CINECA, together with international partners such as the Danish Meteorological Institute, within the MLCast initiative and the EUMETNET E-AI European program.

In this context, IT4LIA AI Factory provided the enabling environment for bringing together and integrating data, models, and high-performance computing infrastructure into a single, coherent operational pipeline serving all stakeholders.

Overall, IT4LIA’s Earth vertical demonstrates how the combination of open data, cloud infrastructure, and artificial intelligence can accelerate the development of predictive solutions in meteorology. The results released are not only scientific advances: they also represent cutting-edge technologies that are finally accessible and scalable for the entire production and industrial system.

Want to find out how to integrate IT-DPC-SRI data into your projects?Visit ArcoDataHub or contact the IT4LIA AI Factory team for more information.