Remote Detection of Nuclear Proliferation

Arms Control Wonk Reader Ben Mayer—a computer science PhD student at the University of Minnesota—attended a talk by William Priedhorsky, Chief Scientist in the International, Space and Response Division at Los Alamos National Laboratory on Remote Detection of Nuclear Proliferation: an Observational Science.

Here are Ben’s notes:

As we have seen from the actions of many countries, in many different parts of the globe, WMD are of great importance to the world over.

• Forgetting the destructive force of nuclear weapons, because most people alive have not seen what they did.
• DoD tends to group many things into the WMD bin. This includes things like Nukes, Chem, Bio, plastic explosives, etc. The problem with this Chemical weapons are difficult to use (cited Tokyo subway attacks, did not mention gas in northern Iraq (don’t know about the amount of technical work needed, volume of gas, etc. that was needed there))

All of the current detections of smuggling were done with good police work or informants. None of them where through technical means. This needs to change.

Of all of the material that we have found, there is a “significant quantity” (as defined by the IAEA, 8kg Pu or 25kg of U235) of material that we know about. We don’t know how much material we don’t know about.

More countries are getting weapons in worrisome parts of the world. E.x. India and Pakistan. These are two countries who have fought wars for the last 25 (?) years, and now they both have nuclear weapons. What types of things this allows to go on in the sidelines is also a problem (A. Q. Khan) NK, Iran also worrisome.

What we really need to do is understand what is happening in the world to have a chance at proper arms control. This involves building models of the world. We can then populate the models from different intelligence sources. At this point one can ask questions of the models to determine if action (or which action) should be taken.

How does one do this? There are signatures and background noise. These are propagated through the environment, distorted, absorbed. These signals can be detected and using computers fused with other data sources. If done properly this fused data can be used to generate knowledge about what is (probably) happening in a location. From this point threat and policy analysis can happen.

There are many observational sciences (astrophysics, paleontology). Astrophysics appears to be very similar to what we want to do here.

Gives several examples of what remote detection is able to do across several domains, including at the WTC site.

In the presented slides there was a good deal of technical information about detecting different types of particles, that is not in the “backup” set.

William talked a great deal about why it is technically difficult to look for many of the different particles that are generated in the process of making suitable nuclear material for ones task (peaceful or not).

“The only reason we found the underground centrifuge facility is because a dissent group held a press conference and we were able to go back and look at satellite images.”

There is a very nice listing of the different methods of how one can detect, with remote sensors, different level of nuclear development. When asked about integration of other information such as aluminum tubes there were two answers. 1. We now know that these more then likely were for making missiles 2. This is the reason that we need a flexible system that can take inputs like that. When talking about this very much implying that the base intelligence that said the tubs were for refinement, was not our work.

Can we automate much of the collection, detection, which detections are interesting, then repeat with new different (type of) detections that can give us more information. One of the things that was discussed over lunch is the need to be able to do many cycles of almost any process, be it detection for nuclear proliferation, or engineering of material goods (humvee armor, etc).

Also can we make (good) enough measurements that we get a better understanding of what is happening in the world?

What William really talked to us about is creating a information management system that many pieces of spatial, temporal, multi-source, multi-frequency data can be collected and analyzed by a person in the context of creating a world view. And wanted to know if the university people had any good inputs. He appeared happy with some of the groups.

In the new slides (this info was not in the presented slides) he has some info about the “nations biggest threat” being the number of doctoral degrees awarded to US citizens in relation to other countries. This is interesting given that the room that he was talking to was ~80% non-US citizens who were academics. At one point he noted that many of the interns at LANL are non-US citizens, and that 300 people who were non-US citizen interns are now working at the lab as citizens and hold high level security clearances.

When asked about the ability of the current system to integrate different sources: the system can have things hacked into it, and most things can be taken into account but it is difficult. This was a very interesting talk because there is a very real problem, and a possible solution inside of computer science.