An algorithm for the detection and tracking of deep convection using satellite data and integer programming / Shishov A.E., Gorlach I.A. // Hydrometeorological Research and Forecasting, 2020, no. 2 (376), pp. 39-59.

Data from geostationary satellites are of great value for monitoring the development of deep convection. However, contemporary methods for their processing and forecasting capabilities are still far from perfect. The objective of the present article is to propose a new convective cloud detection and tracking algorithm based on satellite data. At the first stage of the algorithm, individual deep convective cells or systems are detected by a temperature thresholding technique. At the second stage, clouds detected at two consecutive time steps are matched to estimate their displacement using integer programming. Furthermore, the authors explain the strengths of the proposed method as they compare it to existing alternatives. Its performance is evaluated in a case-study for June 11-12, 2019. During this period, a strong mesoscale convective system was observed over Europe and the European part of Russia. The correspondence of the features of the detected clouds and their displacement to observations suggests that the proposed algorithm is correct.

Keywords: satellite, deep convection, clouds, object detection, optimization, association, integer programming, object tracking

Tab. 1. Fig. 9. Ref. 26.