Storm250-L2: 250 m NEXRAD dataset for ML nowcasting

Immediate prediction of severe storms, known as nowcasting, is one of the most complex challenges in operational meteorology. Current machine learning models rely on high-fidelity radar reflectivity sequences to capture the evolution of convective cells. However, publicly available datasets to date —such as SEVIR, HKO-7, or GridRad-Severe— offer resolutions of 1 to 2 kilometers, smoothing out small-scale structures like cloud walls or mesocyclones, which are crucial for anticipating extreme phenomena. In this context, Storm250-L2 emerges, a storm-focused dataset that leverages NEXRAD Level-II and GridRad-Severe data to provide fixed 250-meter resolution windows, preserving the original polar geometry of the sweeps. Each event includes temporally consistent reflectivity sequences, both per elevation and in a pseudo-composite, packaged in HDF5 tensors with detailed metadata and reproducible manifests. This initiative allows researchers and technology companies to train artificial intelligence models with a granularity never before available in the public domain, improving the ability to anticipate hail, wind gusts, or lightning strikes minutes in advance. Managing the massive data volume —thousands of events over the continental United States— requires scalable infrastructures and custom applications that automate data ingestion, preprocessing, and labeling. This is where companies like Q2BSTUDIO add value: they offer AWS and Azure cloud services to deploy meteorological data pipelines, as well as artificial intelligence solutions for companies wishing to integrate these datasets into early warning systems. Using AI agents, it is possible to develop models that learn storm evolution patterns and update them in real time, while business intelligence tools like Power BI facilitate the visualization of convective activity in interactive dashboards. Cybersecurity also plays an essential role in protecting both raw data and trained models, especially when integrated into critical infrastructures. Q2BSTUDIO, with its expertise in AI for businesses, can assist meteorological organizations and insurance companies in implementing custom nowcasting solutions, transforming high-resolution radar data into operational decisions. The combination of datasets like Storm250-L2 with custom software and cloud platforms opens the door to a new generation of immediate prediction systems, more accurate and reliable, capable of saving lives and reducing material damage.