A REEXAMINATION OF THE PEAK CURRENT CALIBRATION OF THE NATIONAL LIGHTNING DETECTION NETWORK

The peak current calibration of the National Lightning Detection Netwo rk (NLDN) recently reported by Orville (1991) has been reexamined with a roughly threefold larger data set of 57 directly measured stroke pe ak currents, I(Peak) (kiloamperes), and their corresponding NLDN mean normalized magnetic signal strengths, M(peak)BAR (LLP units). These 57 lightning strokes originated in 36 triggered lightning flashes initia ted at the Kennedy Space Center (KSC), Florida, during the research ca mpaigns of 1985-1991. Identification of corresponding I(peak) and M(pe ak)BAR measurements was verified through accurate coincidence in absol ute time of the two independent data sets. The I(peak)-M(peak)BAR data (with one point excluded as an outlier) is apparently linearly relate d with a correlation coefficient of 0.88 1, consistent with that predi cted by application of the transmission line model of the lightning re turn stroke. The regression equation for prediction of I(peak) from NL DN M(peak)BAR measurements is I(peak)=20+0.171M(peak)BAR where the slo pe is expressed in kiloampere/(LLP units). Examination of the overall I(peak)-M(peak)BAR data set for the possible influence of two differen t models of signal strength attenuation with distance, D, (power law, D(beta), and exponential, [exp(alphaD)D]-1) indicates negligible sensi tivity to the proposed variations; other larger error sources Likely m ask the true attenuation effect. Twelve flashes were detected with fou r or more direction finders; a power law fit to the direction finder s ignal strength variation with distance of these individual flashes yie lds a mean beta value of -1.09. Examination of the overall I(peak)-M(p eak)BAR data set for the possible effect of a nonlinear relation betwe en the source stroke peak current and return stroke propagation speed indicates no obvious influence. The 95% confidence bounds for peak cur rent prediction in the mean suggest a percent uncertainty of 10-15% fo r peak currents between 15 and 60 kA. Similar accuracy is expected for peak currents greater than 60 kA provided care is taken to minimize p ossible nonlinear amplification effects in the NLDN data. Below 15 kA, the percent uncertainty rapidly increases suggesting that the inferre d distribution of peak current values less than this limit may be quit e unreliable.