Improving the identification and tracking of mesoscale convective systems over southern Africa using PyFLEXTRKR

This dataset contains tracked mesoscale convective systems over southern Africa (domain: 10°S–40°S, 10°E–60°E), developed using the PyFLEXTRKR tracking algorithm for the period 2019–2024. Two separate tracking runs are included: an original run using the default PyFLEXTRKR configuration and a modified run using adjusted detection parameters. The tracking is performed on gridded satellite data at a spatial resolution of 0.1° × 0.1°. Infrared brightness temperature (Tb) data are obtained from the NASA Global Merged IR V1 dataset, while precipitation information is provided by the Integrated Multi-satellites Retrievals for GPM (IMERG) Final Run, version 07. Both datasets are available at half-hourly temporal resolution. For consistency with IMERG, the IR Tb data are conservatively regridded to 0.1° × 0.1° and merged prior to tracking. In the original PyFLEXTRKR configuration, convective cloud systems are identified by iteratively expanding a contiguous region from a cold cloud core defined by Tb < 225 K, outward to an outer threshold of 241 K, and applying a precipitation rain-rate threshold of 2 mm hr⁻¹, with an additional heavy rain-rate threshold of 10 mm hr⁻¹. In the modified run, detection parameters are adjusted such that the cold cloud core is defined by Tb < 221 K, with expansion to an outer threshold of 235 K, and only a rain-rate threshold of 2 mm hr⁻¹ is applied, with no heavy rain-rate threshold. Tropical cyclones (TCs) are removed from the database using an established methodology, and frontal systems are excluded using the shape complexity index (SCI). For each austral summer season (October - March) and for each tracking run, three CSV files are provided: (1) all systems identified by PyFLEXTRKR, (2) systems remaining after TC removal, and (3) the final set of systems after both TC removal and application of the SCI filter. Jupyter Notebook: A step-by-step Python walkthrough showing how to programmatically interface with the data files. The Notebook Demonstrates: Data Ingestion: Loading the .csv files using the Python pandas library. Basic Exploration: Checking dataset shapes, file metrics, data types, and initial row snapshots (.head()). Descriptive Statistics: Summarising columns and calculating core metrics like means, ranges, and data distributions. Visualisation: Generating basic plots to illustrate foundational patterns within the data.