A Benchmark Dataset for Lightning Nowcasting in Latin America

In 2023, the World Meteorological Organization (WMO) commissioned a pilot project for nowcasting convective weather hazards using artificial intelligence (AI) methods. The project focuses on predicting lightning and quantitative precipitation estimation (QPE) over Latin America, Africa, and southeast Asia. This project is called the AI Nowcasting Pilot Project (AINPP), and is part of WMO's Early Warning for All initiative (EW4All). This benchmark dataset is for evaluating the prediction of lightning in Latin America. The initial dataset consists of 30 days data at 10-minute temporal resolution (5 consecutive days for 6 consecutive months). Contents: mexico_targets.tar.gz: The targets or predictand over Mexico, using 1- and 2-hour accumulations of flash-extent density observed by the GOES-16 Geostationary Lightning Mapper. mexico_example_predictions.tar.gz: Example predictions from the NOAA/CIMSS LightningCast model, matching the files in mexico_targets.tar.gz. south_america_targets.tar.gz: The targets or predictand over South America, using 1- and 2-hour accumulations of flash-extent density observed by the GOES-16 Geostationary Lightning Mapper. south_america_example_predictions.tar.gz: Example predictions from the NOAA/CIMSS LightningCast model, matching the files in south_america_targets.tar.gz.