Multi-scale causal analysis of processes causing lightning and implications to Energy infrastructure

Lightning is a key manifestation of severe convective weather and poses a significant natural hazard to power infrastructure, particularly overhead transmission lines and towers. However, lightning occurrence is governed by the combination of multiple atmospheric and cloud-scale processes. Existing studies largely rely on correlation-based analyses, which provide limited insight into the temporal roles of different precursors prior to lightning.

In this study, we develop an event-driven, multi-scale causal analysis framework based on a large set of real-world lightning events over the UK. Each lightning event is temporally aligned with its preceding atmospheric evolution, combining hourly ERA5 reanalysis variables, including temperature, moisture, and precipitation, with high-temporal-resolution satellite-derived cloud-top height observations. Causal discovery methods are applied to identify lagged relationships at the hourly scale, while robust lag analysis is used to characterise short-timescale cloud-top evolution. The analysis reveals that lightning events are commonly preceded by physically consistent, ordered triggering processes. As a case study, we discuss the implications for power infrastructure risks. The proposed framework provides a data-driven and physically interpretable basis for assessing lightning-related risks to transmission networks and other assets.