|Fig. 1: The familiar mushroom cloud associated with an atmospheric detonation of a nuclear weapon.|
The Comprehensive Nuclear Test Ban Treaty (CTBT) is an international treaty that bans detonation of nuclear weapons by signatory states. The CTBT is not currently in effect, since its implementation has certain requirements that have not yet been met, but monitoring systems for the treaty are already being deployed and studied. Since the CTBT bans all types of detonation, including atmospheric, underground, space and underwater tests, a variety of monitoring systems have to be deployed to provide comprehensive monitoring.
Part of the CTBT is provisions for the International Monitoring System (IMS), a network of different kinds of monitoring stations. The IMS is not fully implemented yet, and its development presents two major challenges: first, covering as much of the earth as possible so that there are not spots where a country could "hide" a detonation, and second, understanding the signals that are seen so nuclear events can be distinguished from earthquakes, other kinds of explosions, meteor strikes, etc. .
The first image that comes to mind with the phrase "nuclear bomb detonation" is the mushroom cloud associated with cold-war era nuclear tests. This is an example of atmospheric detonation. A number of detectable signals accompany such a test, with the two most important being the radionuclides and sound waves that are released into the atmosphere. Radionuclide testing involves filtering air, usually for radioactive isotopes of xenon because they have half-lives of several hours to several days and are chemically inert. This method is challenging because quantities can be dilute, wind patterns can make interpretation difficult, and collection/analysis times should be short to give the greatest sensitivity (you don't want your signal to decay away while you are collecting and analyzing it).  Low-frequency sound (infrasound) waves are also produced in the explosion and propagate around the world, where they can be detected by barometers at one or more of the 60 monitoring stations. The exact likelihood of detection by infrasound varies by location, but there is a projected 90% detection probability of atmospheric detonations above 1 kiloton worldwide, and of detonations above 500 tons on land in the northern hemisphere. 
|Fig. 2: Example shock shapes for a mine collapse and a nuclear detonation. The shapes are very similar, but the nuclear signal has a leading peak from the outward detonation while the collapse signal has a leading trough.|
Underwater explosions are harder to monitor with radionuclide and infrasound because the water damps the signal, but hydrophones (underwater microphones) can detect underwater explosions and underground explosions that happen near the shore. Water can transmit sound waves over long distances, but the terrain of the ocean can damp signals in complicated ways and perhaps more importantly, it can be difficult to distinguish the noise from a nuclear explosions from the noise from, say, an underground earthquake or volcano or a meteor impact. Politics also comes into play here: the treaty has a rather sparse hydrophone system that is prone to blind spots because states do not want submarine exercises to be monitored by the IMS . For these reasons, seismic observations are used as much as possible to corroborate signals from hydrophones .
Seismic detection is also the method of choice for detecting underground detonations, since the ground damps sound waves in the atmosphere and radionuclides can be trapped by rock. Seismic detection methods consist of looking for the pressure and shear waves that result from an underground explosion; again, the primary challenges are having a sensitive and far-reaching system and interpreting the results that you get. For example, a mine collapse and an underground nuclear blast can be distinguished because the initial pressure of a mine collapse is negative (rock getting pulled in by the collapse) whereas a nuclear blast has positive pressure (rock getting pushed out by the blast) . There are also databases containing seismic information about past nuclear explosions that can be used for comparison and calibration , and when India and Pakistan performed underground nuclear tests in May 1998, some of the tests were automatically detected and located seismically within an hour of when they occurred. Like other detection methods much of the challenge of seismic detection is being able to separate signal (nuclear tests) from noise (natural and non-nuclear seismic events), so it is worth noting that the prototype international data center that finds nuclear events was able to distinguish a nuclear test on May 30, 1998 from an earthquake in Afghanistan that occurred 30 minutes before the test. However, India announced some sub-kiloton tests that were not detected, demonstrating how difficult it can be to detect a small test .
A last possible location for nuclear testing is space, and detection of these events is done on a country-by-country basis (space enforcement is not part of the CTBT). This detection would primarily consist of satellite monitoring for the electromagnetic radiation accompanying a nuclear blast. The dearth of data about space tests makes it difficult to quantify how effective the detection methods are, but also presumably testifies to the difficulty of executing a successful and informative space test of a nuclear weapon .
Enforcement is important to the CTBT being taken seriously, so research into and construction of monitoring stations continues today. Detecting all nuclear detonations, and correctly identifying them as such, is a tremendous challenge. So much is at stake with nuclear matters that one must be careful about saying anything too sweeping, since it is likely (certain?) that there's more to this story than what is publicly accessible, but it is probably fair to say that while many nuclear tests can be detected and identified, others, especially small ones, still cannot.
© Katie Malone. The author grants permission to copy, distribute and display this work in unaltered form, with attribution to the author, for noncommercial purposes only. All other rights, including commercial rights, are reserved to the author.
 National Research Council, Research Required to Support Nuclear Test Ban Treaty Monitoring (National Academies Press, 1997).
 T. W. Bowyer et. al., "Automatic Radioxenon Analyzer for CTBT Monitoring," Pacific Northwest National Laboratory, PNNL-11424, November 1996.
 National Academy of Sciences, Technical Issues Related to the Comprehensive Nuclear Test Ban Treaty (National Academies Press, 2002).
 P. G. Richards and W.-Y. Kim, "Monitoring for Nuclear Explosions," Scientific American 300, 77 (2009).
 I. Bondar et al., "Location Calibration Data for CTBT Monitoring at the Prototype International Data Center," Pure Appl. Geophys. 158, 19 (2001).
 B. Barker et al., "Monitoring Nuclear Tests," Science 281, 1967 (1998).