The apparent yield of a nuclear test is estimated from the magnitude which, in turn, is estimated from the amplitude of the waveform arriving at a seismic station. Below is a brief overview of the calculation.
There are several magnitude scales. Richter Magnitude was the first, there is also body wave magnitude (mb) and surface wave magnitude (Ms). They are all estimated in a similar fashion and as such the values obtained are very similar.
For a specific event the magnitude is calculated for each seismogram arriving at each station. An average is then obtained using estimates from each station in the network. The magnitude estimates from different networks often vary slightly as different groupings of stations were used in each calculation.
Yield is calculated from magnitude using the following equation
where mb is body wave magnitude, Y is yield in kilotons. A and B are constants, dependent on the geology local to the test. The constants used by the U.S. government are classified, however seismologists have published various estimates:
As the exact value of A and B appropriate for any particular test site are unknown, we estimate the range of possible yields using a reasonable range of values for A and B.
There are many postulated evasion scenarios in which countries who wish to conduct a nuclear test use some strategy to prevent the global seismic network from detecting the event. One of the more serious strategies is decoupling. The energy from the blast is decoupled from the earth to some extent to reduce the apparent size of the blast -- perhaps below the detection threshold. This is done by detonating the device in a large underground cavern so the device is not in direct contact with the surrounding rock. This approach could reduce the apparent yield by up to a factor of 70.