Lightning drills a 1 to 2 cm hole in the sky several kilometers long. Thunder is what we hear when the atmosphere is snapped open and then refills the hole. Imagine the sound when a hole about 1 meter in size gets drilled through tens of kilometers in a second or less.
That's what happens when a significant meteor comes down to within 50 km of the surface. The sound produced by this shockwave (particularly the infrasound below 20 Hz) can be detected from several thousand kilometers away. These events are of interest not only because of the spectacular display and their effect on unwary populations (on September 27 an event over the state of Orissa in eastern India caused several people to collapse in shock, and one person died from a possibly connected heart attack), but also because they are detonations in the atmosphere, and could be mistaken for nuclear tests. In fact, the energy of these bolides is often quoted in kilotons of TNT, the standard energy equivalent for nuclear shots. A bolide with total energies over 1 kt is needed for infrasound detection with the current comprehensive test ban treaty infrasound international monitoring system, and estimates of the bolide flux in this range are about 10 events worldwide per year. About 6 to 7% of a nuclear blast's energy goes into sound, and the estimates for bolides range from 0.2 to about 7%.
Here's a great summary of the collision flux with Earth, ranging over 14 orders of magnitude, from an often quoted letter to Nature by Brown, Spalding, ReVelle, Tagliaferri and Worden (10.1038/nature01238 - PDF file):
Other methods that are used to detect these events are satellite optical and IR observations of the fireball, radio reflections, seismic signals (again, just a form of sound), and ground-based video. The proliferation of security cameras has in fact been a good source for fireball data, but often what is seen is a reflection off a car or a window, since the cameras are looking down to catch crime, and not up to catch celestial phenomena.
Things to remember: shooting stars and fireballs are neither stars, fire, nor comets; groundfall (pieces reaching the ground) from an event is extremely rare and very difficult to find, but are often mistakenly reported; eyewitness accounts are almost without exception useless for research (we are poor observers of exciting events).
There is a great article by Alan Hildebrand of the Royal Astronomical Society of Canada here that gives a nice mathematical summary of the chances of seeing a meteorite fall. Summary: of the approximately 7,000 groundfalls in a year, about 5 have a chance of being seen. His comment about a herd of dairy cattle extending a person's awareness cross-section made me laugh...
There is a nice page about the 1947 Sikhote-Alin fireball (and groundfall!) here.
A Colorado website tracks fireballs, here, and seems to have a scientific approach.
Then again, if you are listening on Christmas Eve, and hear a rumble under clear skies, it could be the crack of a reindeer whip you hear.
