Understanding the RF emissions from a spark

Another question on RF interference.

I understand sparks cause serious RF interference. But I'm trying to understand the physics. From my understanding, an arc of several thousand volts causes ionization. But why does ionization generate rather powerful electromagnetic waves?

To add to my confusion, I found this site that states "The frequencies are random and come and go in fractions of a second". Mathematically speaking, I'm uncertain about the term "random".
Is it truly "random", and if so, across which frequencies?

Alternately I considered the problem from a Fourier perspective. A spark causes a rapid voltage change in a very short time. I considered essentially a very short pulse. A very short pulse in the time domain is correlary to a sinc function in the frequency domain. Or in other terms, that pulse includes MANY sine/cosine harmonics to generate. These harmonics are what cause interference?

The ionization is not the cause of the RFI. It's just a another result of the arc current, which is the real perpetrator.

You're on the right track with regards to its spectral content. A spark is always a brief, transient event, and therefore it has a very broad spectrum. That's why it's capable of affecting pretty much any type of equipment. It's not really random, but it's dependent on so many factors (humidity, arc length, total charge, air pollution, etc) that there's really no way to predict its exact behavior.

Wow great response mtwieg. Thanks!

The spark is an impulse function. A real world spark has some rise time which cause the energy to drop off at higher frequency.

The first radio used this for the exciter which was an electromagnetic vibrator creating spark impulses into a resonant tank circuit. Just like a Tesla coil. The principle is the same as a frequency multiplier.

It was the type of transmitter used on the 'Titanic'.

To a better understanding of what a spark generates:
a. in the early radio (even Mr. Hertz used the spark gap) to generate the electromagnetic wave, a dipole radiator (or similar structure) connected to a spark gap can radiate a wave when there was a high-voltage discharge through the spark gap. The principle behind it is that the charges in the radiator were abruptly moved and this generated the electromagnetic wave. The dipole and other radiators were tuned to resonate at a selected frequency for the best system performance. The spectrum still was quite wide but the resonance narrowed it. Recently, UWB technology utilizes non-resonant wideband spectra generated similarly but instead of a spark gap, semiconductor or plasma switches are used.
b. The spark-gap discharge alone is a fast charge movement; it has a wide spectrum extending from UV over visible, IR and even into microwave region. So far it has not much use in practice, for communication, etc.