Reliability of Explosive Detection Dogs

By Jerry Bradshaw

Recent reports surrounding the Christmas day bomb attempt on a Northwest Airlines Flight from the Netherlands by Umar Farouk Abdulmutallab and then 2 subsequent events have raised questions about detection devices, both K9 and machine based systems. In one of the subsequent cases, a K9 alerted to a bag in Minneapolis which contained no explosive compounds upon further inspection, but shut down the terminal for some time, inconveniencing travelers. In another case, a swab machine in California sounded an alarm for high explosives on a bag that ostensibly contained only jars of honey, once again causing terminal closure and further inconvenience.

Now, just because a dog alerts on a suitcase and no bomb is found, doesn’t mean that the dog is incorrect. I found a number of popular press articles when I typed “bomb dog reliability” into Google, pointing fingers at seemingly unreliably canines. First one incident is good news but it doesn’t make a statistical case. Second, laypeople with no understanding of the process by which detection dogs are trained would, seemingly reasonably, make such a determination. However it is a false choice. Here are some things to consider as this debate continues:

  1. No detection system is infallible. False positives and false negatives occur all the time. Two doctors with years of training can look at the same diagnostic x-ray and make 2 different diagnoses. That doesn’t mean we should abandon doctors as the vehicle for diagnosing medical conditions.
  2. Machine detection devices for explosive detection often rely on human interpretation (x-rays in use now for example) and the human element can both miss a target device (false negative) or cause an alert to a harmless device (false positive).  A swab machine in California that recently detected the presence of TNT and PETN, an explosives compound normally called “pentolite,” however  the bag itself contained only honey. TNT can have a yellow color. Honey is kind of close. Not a bad call to close things down. Better safe than sorry. Machines can make mistakes too.
  3. Full body scanners (modern x-ray vision) can be easily defeated by hiding explosives in a body cavity. However, even in such a situation, residual odors are likely present for a dog to detect.
  4. Explosives dogs are usually certified to perform at a 90% + level of reliability. Perfection is attained in certification all the time, however, in a number of repeated trials, dogs will occasionally make a mistake. In most cases during certification, we penalize false negatives (missing a device) more highly than false positives (alerting when no device is present).
  5. Explosives dogs are trained to detect chemical compounds, some of which have very closely related relatives in terms of their molecular structure that are completely harmless. For example some grease compounds can contain nitrate ions that are harmless in a grease compound, but are deadly in a smokeless powder pipe bomb. Better to get a false positive, and rule out anything harmful.
  6. Many dogs will react to very low vapor pressures, or lingering odors. These are odors that are there, but you can’t see the source. It is like if I pop some popcorn in your house, then leave with the bag and eat it outside. You will come home and smell popcorn, but there won’t be any for you to see. More to the point, suppose a person reloads their own deer hunting ammunition, and their suitcase is nearby, or they touch it after performing the reloading. Residual amounts of smokeless powder would be detected by a dog – they can detect in the range of parts per trillion (a detection limit no machine in existence can yet match) – however, there would be no bomb in evidence when the bag was searched. The dog is not wrong, though no bomb was found.
  7. Security, for anything, must be conducted in a layered approach. By layering canine and machine detection together the probability that a substance will get past the system drops precipitously. Canines are used in many settings to detect substances on people, for example in searching visitors and inmates in prison for contraband items including drugs, tobacco and cellular telephones. This technology and procedure can be adapted to airports in my opinion as part of a properly layered approach.
  8. Americans are believers in high technology. However, machines are expensive and not without their drawbacks. Canine systems are in use in other countries, especially in Europe, in a much greater proportion, and we should consider layering both low and high technology systems together, in addition to proper intelligence, eliminating bureaucracy with respect to watch lists to maximize the likelihood that a terrorist would be foiled well before getting on an airplane.

Politicians will likely call for more machines, because when people see machines, they feel safer. Let’s concentrate on really making people safer. Ignorance is not always bliss.