The ATRACKCS Algorithm for tracking convective systems (Robledo et al., 2024) is an open-access Python tool designed for the detection and tracking of Mesoscale Convective Systems (MCSs). The methodology used in MCSMIP for MCS detection and characterization follows these steps:
- Initial Detection: At each time step, the algorithm identifies cold cloud systems based on a predefined threshold for brightness temperature (Tb ≤ 241 K). These systems are then converted into polygons, representing the areas where cold cloud tops are observed.
- Geographical Transformation and Area Filtering: The identified polygons are transformed from geographic to plane coordinates. Then, the algorithm calculates the approximate area of each polygon. Regions with an area of 5000 km² or less are discarded.
- Polygon Buffering and Merging: A buffer of 0.1 degrees is applied around each polygon, and overlapping polygons are merged. This ensures that contiguous systems are treated as a single entity.
- Precipitation Thresholding: The precipitation attributes of each identified system are estimated. Any system that fails to meet a minimum precipitation rate of 4 mm h⁻¹ in at least five pixels is excluded from further analysis.
- Tracking and Linking: To link polygons across time steps, ATRACKCS uses a threshold for the minimum overlapping area between consecutive time steps. A 25% overlap in area is the minimum required to associate the polygons with the same system across time. In cases where multiple systems overlap at a given time step, ATRACKCS evaluates which system meets the criteria best, selecting the system with the highest overlap with the previous time step's track. Systems that do not meet the criteria for continuation are treated as new systems if they meet the detection thresholds.