An automated technique and decision support system for lightning early warning

This paper presents an innovative approach to enhancing lightning early warning systems through a quadrant-based monitoring technique. By segmenting the observation area into quadrants, this study evaluates various radius configurations to identify the most effective warning area for minimizing false alarm rates (FAR), failure to warn (FTW), operational downtime, and lead time. The multicriteria method TOPSIS is utilized to identify the best warning area that aligns with the preferences of decision-makers. The methodology was tested in two regions in Brazil, in the Southeast (P1) and in the North (P2), whose climatic features are very different, and both are characterized by mining activities and vulnerability to lightning threats. The quadrant-based monitoring approach demonstrates competitive results over the concentric circle and machine learning-based models, offering a simpler, more interpretable, and maintainable solution. The solutions found for P1 showed, on average, a 0.519706 FAR rate, 0.160438 FTW rate, 0.025319 operational downtime, and 0.219049 lead time. For P2, the FAR rate was 0.624436, the FTW rate was 0.253778, operational downtime was 0.027488, and lead time was 0.249846. A sensitivity analysis was conducted to confirm that the proposed approach acts as a local search in the vicinity of the area monitored by the concentric circle. The findings indicate that the proposed method enhances the safety of outdoor employees by providing more accurate and timely lightning warnings. Additionally, this research presents advancement of meteorological forecasting and occupational safety, underscoring the potential for integrating automated systems and multicriteria decision-making in real-time operational settings.