Hailstorm Identification and Tracking over Brazil (HIToB): A Storm Polygons Database From GOES ABI Data from 2018 to 2023

This dataset comprises a detailed record of deep convective storm events tracked across South America from 2018 to 2023, utilizing brightness temperature (BT) data from Channel 13 of the GOES-16 Advanced Baseline Imager (ABI) and the TATHU (Tracking and Analysis of Thunderstorms) toolset, that caused hail-fall over Brazil. The database includes storm identification, tracking details, and associated meteorological variables such as brightness temperature statistics inside the storm polygon at each scene and event classifications (e.g., spontaneous generation, continuity, split, merge). The storms were detected and tracked based on brightness temperature threshold of 235 K, with tracking data refined by a 10% overlap criterion between sequential scenes. The tracked convective systems were filtered for intersections in space and time with verified hail reports from Prevots group. The whole family of storm polygons that matched the reports were exported to this database with SpatiaLite enabled dtaa format, in order to make it easier for spatial data queries and analysis. Some example queries using Python library SQLAlchemy are displayed in the code repository as well as the process of creating the tables in the database.The data is organized in three tables: "storms", "storm_events" and "intersections". In table "storms" are the records of storm families identifier. Each identifier represents a sequence of storm polygons tracked over subsequent satellite scenes. Table "storm_events" holds the evolution of the storm's geometry through its lifecycle, including BT's mean, minimum and standard deviation inside the storm polygon; as well as storm's pixel count (i.e. storm size). Intersections table stores every instance where a storm event polygon intersects with a hailstorm report's buffer at the corresponding time. In total, there are 9893 intersections belonging to 2172 unique storm families.本数据集详细记录了2018年至2023年间南美洲境内追踪得到的深对流风暴事件，所用数据源为GOES-16先进基线成像仪（Advanced Baseline Imager, ABI）13通道的亮温（brightness temperature, BT）数据，以及雷暴追踪与分析工具集（Tracking and Analysis of Thunderstorms, TATHU），所追踪的风暴均为在巴西境内引发冰雹灾害的事件。该数据库包含风暴标识、追踪详情，以及关联气象变量，例如各卫星影像场景下风暴多边形内部的亮温统计特征，还有事件分类类型（如自发生成、持续发展、分裂、合并等）。风暴的检测与追踪基于235K的亮温阈值，且通过相邻卫星影像场景间10%的重叠准则对追踪结果进行了优化。所追踪的对流系统经过筛选，仅保留与普雷沃茨（Prevots）团队验证过的冰雹报告在时空上存在交集的事件。所有匹配报告的风暴多边形集合均以启用空间扩展的SpatiaLite数据格式导出至本数据库，以简化空间数据的查询与分析工作。代码仓库中还展示了使用Python库SQLAlchemy编写的示例查询语句，以及数据库表的创建流程。本数据集共分为三张数据表："storms", "storm_events"与"intersections"。其中"storms"表存储风暴族标识记录，每个标识对应后续卫星影像场景中追踪得到的一系列风暴多边形序列。"storm_events"表记录了风暴在整个生命周期中的几何演化信息，包括风暴多边形内部的亮温均值、最小值与标准差，以及风暴的像素计数（即风暴规模）。"intersections"表则存储了所有风暴事件多边形与对应时间点冰雹报告缓冲区发生时空交集的实例。总计包含9893个时空交集实例，隶属于2172个唯一风暴族。