Threshold charge moment changes for sprite initiation and lightning-to-sprite delays

Measuring the lightning charge moment change required to initiate sprites helps to understand the mechanism that generates these mesospheric optical emissions and their effect on this region of the atmosphere. The transient magnetic field radiated by lightning discharges was continuously measured remotely by using a wide band ELF (Extremely Low Frequency) receiver at Duke University during the summer of 2000. Optical emissions from sprites were directly monitored on video recordings made at Yucca Ridge Field Station, Colorado, during the STEPS (Severe Thunderstorm Electrification and Precipitation Study) Campaign in May, June, and July 2000. By using a deconvolution method based on a mode-theory electromagnetic propagation model, the current moment and charge moment of the recorded sprite-associated lightning discharges were calculated. The high time resolution of our measurements enables us to accurately measure the charge moment change at the time of sprite initiation, which is the goal of this study. The current and charge moment waveforms of the 881 lightning discharges accompanied by sprites during the STEPS campaign of 2000 were extracted. We analyze in detail only those sprite events with clear sprite currents or with short lightning-to-sprite delays (<6 ms) because of the sprite onset time uncertainty from the video images. Data analysis shows the threshold charge moment change for sprite initiation varies from 123 (C-km) to 2470 (C-km). Also we investigated the experimental data supporting the theoretical prediction that lightning discharges associated with high speed of charge removal can produce sprites with small charge moment change. The frequency of mesospheric sprite current occurrence is found to vary from 0 to 42% from day to day, with an average of 8.7% over the 17 days studied. Sixty-five sprites associated with sprite currents were analyzed and the statistical distribution of these charge moment changes is given. Also the statistical distribution of lightning-to-sprite delays is given with an estimated average of 36.8ms. The deduced threshold charge moment change is compared with theoretical expectations.