Cave Radio. Low frequency radio signals can penetrate the earth for several hundred feet and this makes it possible to communicate between people working the surface and people inside a cave. The low frequencies involved require that the signal be transmitted and received using coils of wire instead of ordinary antennas.

It is possible to use a cave radio to measure the GPS position of stations in the cave. You just move the receiving coil around above ground until you find the location of the maximum signal. In theory, this point should be right above the transmitting coil. Once you've located the point of maximum signal strength, you can take a GPS reading to know the East and North position of the cave location in Geographic Coordinates.

You can also measure the depth of the station, by moving a short distance from the location and measuring the angle of magnetic fields. You can then use Trigonometery to calculate the depth. Here are some links the give more details on the process:

 1) Radio Location - An Introductions. 2) Radio Location in Caves

 Radio Location Accuracy. Many people assume that since you detect a cave location by transmitting a signal directly to the surface, the measurements must be accurate. Depending on circumstances, these measurements may not be any more accurate than compass-and-tape surveys. 

Depth Measurements. Because measuring depth requires measuring the angle of the magnetic field, it is widely understood that depth measurements with cave radios is not very accurate. This is especially true with deeper caves because the deeper the signal source, the more effect small errors can have.

X-Y Measurements. As I pointed above, most people assume that the X-Y or East-North aspect of a radio location will be more accurate than the depth measurement. However, similar problems apply to X,Y aspect of a radio location. There are several sources of error for the X/Y aspect of Radio locations:

1. Coil Leveling. In order for the X,Y aspect to be accurate, the transmitting coil must be absolutely horizontal (level) and the surface receiving antenna absolutely vertical (orthogonal). Most cave radios use a Spirit Level to make the coils horizontal. Depending on the curve of the glass in the Spirit Level, the accuracy can be as little as 2 degrees per millimeter of bubble displacement. (Here is some information on Spirit Level Sensitivity.)

In addition, the spirit level must be attached to the coil so it is at a perfect right angle to the plane of the coil. Cave radios are generally home-made devices and so they lack the precision of a factory-made product. Even if they were factory-made, that wouldn't guarantee any higher accuracy than the typical instruments we use for regular surveying. For example, even a precision-made product like a Suunto Inclinometer only has a resolution of 1 Degree. That means it very unlikely that a cave radio could leveled more precisely than 2 degrees. A 2-degree error in level would produce position errors of 16 feet at 500 feet of depth.

Take a look at the image at this link: Image Of Cave Radio Coil. As you can see, it would be difficult to accurately level a transmitter coil like this.

Finally, just like other cave instruments, spirit levels are subject to the same type of reading errors that compasses and inclinometers are, especially in the harsh cave environment. Judging by the number of blunders I see in ordinary cave survey measurement, it wouldn't be surprised if you had large numbers of leveling error attributable to human error.

2. Coil Manufacture. Even if the coils are perfectly horizontal, distortions in the magnetic field can be produced by the way the coils are manufactured. The coil itself should be symmetrical and as flat as possible, without any folds or bumps, or the field may be distorted. Again, home-made coils are not likely to be flat, especially if the coils are made to fold for easy transport through the cave.

3. Rock Characteristics. In addition, the rock above and around the cave may affect the magnetic field. Most limestone deposits have pyrite and limonite deposits that can distort the magnetic field. This is particularly true around joints and faults were caves are concentrated and ground water has the most chance to redeposit minerals. Finally, changes in the saturation of ground water can affect the magnet field. As an example of this type of phenomena, there are large magnetic distortions in the magnetic field in Lechuguilla cave near the "Rift" area. I also have an aviation map that warns of large magnetic compass deflection near Manitou Springs, Colorado, which is one of our favorite caving areas.

This is what Ian Drummond said about the magnet field distortions:

"Unfortunately the effect of the conductive ground is not just to weaken the intensity of the magnetic field, but also to induce a component which is out of phase. As a result the magnetic field on the surface above the cave is no longer plane-polarized, but rotating."

http://caves.org/section/commelect/drupal/files/Speleonics/splncs07.pdf

If the magnetic field is no longer "plane-polarized" then peak position of the radio signal is likely to be offset from the actual underground position of the transmitter.

How To Enter Cave Radio Locations In Compass. In many ways, radio locations are subject to the same types of errors that ordinary shots are subject to. For this reason, it is best to enter them like survey shots. That means that you shouldn't enter a fixed station with the X,Y taken from the GPS and the Z derived by subtracting the depth.

What you should do is enter a fixed GPS location for the place on the surface where the radio location was taken. Then you add an artificial vertical shot with an inclination of -90 and a length equal to the radio depth measurement. Compass would then treat it as an ordinary survey shot with the same error processing as a regular shot.

GPS Accuracy. Most people assume that GPS receivers are highly accurate, but even they have their own source of errors. For example, non-"survey grade" GPS receivers can have errors as much as 10 meters or 30 feet. Vertical errors can be has much as 20 meters or 60 feet. (Thanks to Paul Jorgenson these numbers.)

Here is a summary of the types of errors with radio locations:

http://www.chaos.org.uk/survex/cp/CP10/CPoint10.htm