A new instrument has been developed and deployed for sensitive reception of broadband extremely low frequency (ELF) (defined in this paper as 300-3000 Hz) and very low frequency (VLF) (defined in this paper as 3-30 kHz) radio signals from natural and man-made sources, based on designs used for decades at Stanford University. We describe the performance characteristics of the Atmospheric Weather Electromagnetic System for Observation, Modeling, and Education (AWESOME) instrument, including sensitivity, frequency and phase response, timing accuracy, and cross modulation. We also describe a broad range of scientific applications that use AWESOME ELF/VLF data involving measurements of both subionospherically and magnetospherically propagating signals.

We present the first global distribution of the average estimated peak currents in negative lightning flashes using 1 year of continuous data from the Vaisala global lightning data set GLD360. The data set, composed of 353 million flashes, was compared with the National Lightning Detection Network^TM for peak current accuracy, location accuracy, and detection efficiency. The validation results demonstrated a mean (geometric mean) peak current magnitude error of 21% (6%), a median location accuracy of 2.5 km, and a relative ground flash detection efficiency of 57% averaged over all positive and negative reference flashes, and 67% for all reference flashes above 15 kA. The distribution of peak currents for negative flashes shifts to higher magnitudes over the ocean. Three case study 10°×10° regions are analyzed, in which the peak current enhancement is extremely sharp at the coastline, suggesting that the higher peak currents for oceanic lightning cannot be solely attributable to network artifacts such as detection efficiency and peak current estimation error. In these regions, the geometric mean and 95th percentile of the peak current distribution for negative cloud to ocean flashes is 22%–88% and 65%–121% higher, respectively, compared to cloud to ground flashes in nearby land regions. Globally, the majority of all negative flashes with estimated peak current magnitude above 75 kA occur over the ocean.

Very Low Frequency (VLF, 3–30 kHz) radio waves emitted from ground sources (transmitters and lightning) strongly impact the radiation belts, driving electron precipitation via whistler‐electron gyroresonance, and contributing to the formation of the slot region. However, calculations of the global impacts of VLF waves are based on models of trans‐ionospheric propagation to calculate the VLF energy reaching the magnetosphere. Limited comparisons of these models to individual satellite passes have found that the models may significantly (by >20 dB) overestimate amplitudes of ground based VLF transmitters in the magnetosphere. To form a much more complete empirical picture of VLF transmitter energy reaching the magnetosphere, we present observations of the radiation pattern from a number of ground‐based VLF transmitters by averaging six years of data from the DEMETER satellite. We divide the slice at ∼700 km altitude above a transmitter into pixels and calculate the average field for all satellite passes through each pixel. There are enough data to see 25 km features in the radiation pattern, including the modal interference of the subionospheric signal mapped upwards. Using these data, we deduce the first empirical measure of the radiated power into the magnetosphere from these transmitters, for both daytime and nighttime, and at both the overhead and geomagnetically conjugate region. We find no detectable variation of signal intensity with geomagnetic conditions at low and mid latitudes (L < 2.6). We also present evidence of ionospheric heating by one VLF transmitter which modifies the trans‐ionospheric absorption of signals from other transmitters passing through the heated region.

In this research, we present data-driven forecasting of ionospheric Total Electron Content (TEC) using the Long-Short Term Memory (LSTM) deep recurrent neural network method. The random forest machine learning method was used to perform a regression analysis and estimate the variable importance of the input parameters. The input data are obtained from satellite and ground based measurements characterizing the solar-terrestrial environment. We estimate the relative importance of 34 different parameters, including the solar flux, solar wind density and speed, the three components of interplanetary magnetic field, Lyman-alpha, the Kp, Dst and Polar Cap (PC) indices. The TEC measurements are taken with 15-second cadence from an equatorial GPS station located at Bogota, Columbia (4.7110° N, 74.0721° W). The 2008 - 2017 data set, including the five top parameters estimated using the random forest, is used for training the machine learning models, and the 2018 data set is used for independent testing of the LSTM forecasting. The LSTM method as applied to forecast the TEC up to 5 hours ahead, with 30-minute cadence. The results indicate that very good forecasts with low RMS error (high correlation) can be made in the near future and the RMS errors increase as we forecast further into the future. The data sources are satellite and ground based measurements characterizing the solar-terrestrial environment.

Analysis of ELF/VLF broadband data from Palmer Station, Antarctica indicates that 76% Terrestrial Gamma‐ray Flashes (TGFs) detected on the RHESSI spacecraft occur in association with lightning‐generated radio atmospherics arriving from near the footprint of RHESSI and within a few ms of the TGF. The remaining TGFs are not associated with any radio atmospheric, thus by implication CG lightning. The peak currents of TGF‐associated lightning discharges are often among the most intense from a given storm, with the degree of this association apparently varying between oceanic and land regions. The time‐integrated ELF energy of the associated sferics (and thus the lightning charge moment) exhibit much less tendency to be large. Statistical analysis of the spread in arrival time suggests a ∼2 ms variance due to factors other than geometry and measurement error.

Accurate models of trans‐ionospheric propagation are needed to assess the role of Earth‐originating very low frequency (VLF) electromagnetic waves in radiation belt dynamics. Recent studies have called the relatively crude early trans‐ionospheric models into question, finding that they underestimate the attenuation by 20–100 dB. A full wave model that includes all of the relevant physics has recently become available and experimentally verified to within a few decibels via comparison to more extensive satellite data. Using this model, we discuss the importance of wave polarization, incidence angle, bearing, ground conductivity, horizontal distance from the source, and the ionospheric profile, all of which are demonstrated to play a significant role in the trans‐ionospheric propagation. trans‐ionospheric attenuation estimates are provided both for the case of vertical incidence of a whistler‐mode wave and for the case of magnetospheric injection from a dipolar terrestrial VLF source. These estimates agree with observation to within ±6 dB. The remaining discrepancy may be attributable to ionospheric variation and/or factors not captured by our horizontally stratified model. On the basis of the full wave treatment presented herein, we find that the earlier work showing a >20 dB overestimation by traditional models results from the unrealistic simplifying assumption that the wave is vertically incident onto the ionosphere, exacerbated by the fact that most of the satellite data used for comparison came at large horizontal distances (hundreds of kilometers) from the source.