Time-domain radio, aka ultrawideband or pulse radio, is mostly considered in the context of very-low-power communication. But there's an interesting very-high-power physical phenomenon that could be useful for time-domain radio: lightning. The mechanism that triggers lightning is not well understood; when it happens, the electric field is still two orders of magnitude below the voltage needed to ionize homogeneous air. Then, the current jumps from effectively zero to some tens of kA within a few microseconds, with a conductor length of typically a few kilometers. This transmits a powerful electromagnetic impulse horizontally across the bands up to a few hundred kHz, an impulse powerful enough to be detected at considerable distance. If we figure on an antenna impedance of 63 ohms, 30 kA will generate a radio pulse of about 60 gigawatts.
How far away can you detect a 60-gigawatt pulse from? If we assume a detection threshold of -70dBm, which works fine in Wi-Fi cards, and a ten-square-meter dish antenna, you can detect a signal down to 10⁻¹¹ W/m², so you hit that threshold when the power is spread out over 6×10²¹ m², which is when the pulse has expanded into a sphere 22 billion meters in radius. This is about a sixth of the distance from the Earth to the Sun. If you can detect down to -85dBm, you can apparently detect the lightning strike from just about anywhere in the inner solar system.
So it seems safe to speculate that you should be able to detect a lightning strike using shortwave radio from anywhere on Earth, and using other radio bands, perhaps from some of the other inner planets too. It really seems like overkill for terrestrial communication.
You still can't make lightning from scratch, but you can trigger it. The traditional approach is to blaspheme, but that has been shown both experimentally and theoretically to have very low efficacy; additionally, the large amount of time required to blaspheme (hundreds of milliseconds or more) would dramatically limit your possible communications bandwidth. More modern approaches, shown to work fairly reliably, include firing rockets into clouds trailing wire from spools, and ionizing paths in air using ultraviolet lasers. Presumably particle accelerators can also ionize paths in air. (I've seen speculation that cosmic-ray strikes play a role in creating lightning leaders.)
A problem with laser-induced ionization is that ionized air is opaque, so it tends to be self-limiting; nevertheless, some experiments have found success using this method.
Particle accelerators might work better; particle beams are attenuated by air, but no more by plasma than by regular air, less so even. And you could maybe use the particle accelerator to produce a beam of synchrotron radiation or X-rays to form the ionized path, rather than firing subluminal particle beams into the air. Alternatively you could perhaps use an X-ray laser.
You probably can't find a place where you can trigger lightning more often than a few times a month, and you can probably only control the triggering to a precision of a microsecond or so within a window of a few seconds, so your total system bandwidth isn't going to be more than a few dozen bits per month.
This approach to radio communication has the advantage that you could transmit data over interplanetary distances without being detectable as a radio source without access to the spreading code.