Automated analysis of lightning leader speed, local flash rates and electric charge structure in thunderstorms

A Lightning Mapping Array (LMA) maps radio pulses emitted by lightning leaders, displaying lightning flash development in the cloud in three dimensions. Since the last 10 years about a dozen of these advanced systems have become operational in the United States and in Europe, often with the purpose of severe weather monitoring or lightning research. We introduce new methods for the analysis of complex three-dimensional lightning data produced by LMAs and illustrate them by cases of a mid-latitude severe weather producing thunderstorm and a tropical thunderstorm in Colombia. The method is based on the characteristics of bidrectional leader development as observed in LMA data (van der Velde and Montanyà, 2013, JGR-Atmospheres), where mapped positive leaders were found to propagate at characteristic speeds around 2 · 104 m s-1, while negative leaders typically propagate at speeds around 105 m s-1. Here, we determine leader speed for every 1.5 x 1.5 x 0.75 km grid box in 3 ms time steps, using two time intervals (e.g., 9 ms and 27 ms) and circles (4.5 km and 2.5 km wide) in which a robust Theil-Sen fitting of the slope is performed for fast and slow leaders. The two are then merged such that important speed characteristics are optimally maintained in negative and positive leaders, and labeled with positive or negative polarity according to the resulting velocity. The method also counts how often leaders from a lightning flash initiate or pass through each grid box. This "local flash rate" may be used in severe thunderstorm or NOx production studies and shall be more meaningful than LMA source density which is biased by the detection efficiency. Additionally, in each grid box the median x, y and z components of the leader propagation vectors of all flashes result in a 3D vector grid which can be compared to vectors in numerical models of leader propagation in response to cloud charge structure. Finally, the charge region altitudes, thickness and rates are summarized from vertical profiles of positive and negative leader rates where these exceed their 7-point averaged profiles. The summarized data can be used to follow charge structure evolution over time, and will be useful for climatological studies and statistical comparison against the parameters of the meteorological environment of storms.