How NoSQL Is Helping Allay Seattle’s Radiation Fears

Researchers at the University of Washington utilized the Cloudant NoSQL database as part of an experiment that determined radiation levels in Seattle as a result of the recent Fukushima nuclear disaster are “well below alarming limits.” The research team, which includes Cloudant Founder and Chief Scientist Mike Miller (his other title is research associate professor of physics), studied particles captured from the five air filters at the university’s Physics and Astronomy building and used Cloudant’s CouchDB-based BigCouch database to store and process the data from its experiments.

The research team's shielded germanium detector and data acquisition hardware.

The team’s results show radioactive particles first reached Seattle on March 17, but they were less abundant in volume and type than was expected based on research carried out after the Chernobyl disaster in 1986. In fact, the project’s official website states, “We stress that the overall amount of the radioactivity is extremely low, at least thousands of times below EPA limits.” Miller attributes the difference in radiation levels resulting from Chernobyl to the fact that the reactor at Chernobyl “was operating at full steam when it exploded,” whereas his team’s research suggests the Fukushima plant’s automatic safety system must have kicked in when the earthquake hit, turning off the reactor in the process.
Miller’s association with Cloudant certainly influenced his team’s decision to use the product, but BigCouch — which is more big-data-focused than its web-application-focused NoSQL counterparts — is particularly well-suited for this type of job. According to Miller, the team created “a MacGuyver-like setup” in order to start monitoring radiation levels in a hurry. After installing the special air filters and sampling “gigantic quantities of air,” the trapped particles were run through a germanium detector to determine the unique fingerprints of isotopes created by nuclear fission. Miller says the general belief is that radioactive isotopes travel through the air attached to dust particles or concrete particles from the explosion.
Cloudant’s BigCouch database let the team keep up with a steady flow of data so it could processĀ  and analyze it, then share it with the various stakeholders in near-real-time. The team was changing the data about 20 times per day and writing complex workflows to process it, two tasks that fall into BigCouch’s wheelhouse. The database has a built-in MapReduce engine to enable writing and processing the workflows, and it allows for secondary indices, which users can populate with new data from their MapReduce jobs and query very quickly. Miller actually helped create Cloudant in 2008 while doing post-doctoral research at MIT that involved analyzing huge data sets from CERN’s Large Hadron Collider.
Although humans might not be at risk from the Fukushima-based radiation, their scientific experiments might not be so lucky. In an article by the University of Washington’s Office of News and Information, Miller’s colleague R.G. Hamish Robertson explained that the radiation levels “can raise havoc with sensitive physics experiments. That includes one called Majorana, in which the UW physicists are deeply involved, that is being planned for a lab nearly [one] mile down in the proposed Deep Underground Science and Engineering Laboratory in the old Homestake Mine in Lead, S.D.” Miller told me that project, which involves trying to determine what the universe is made of, has been suspended because the next step is to bring highly sensitive homegrown copper up from underground for machining, but that even the trace levels of radiation present in South Dakota could taint it and ruin the experiment.
The currently-available paper sharing the team’s results and insights mentions some uncertainty based on questions about the particle sizes they were trying to capture, but Miller said his team has made progress on this front and that they’re “very confident” in the results. The team will release a final paper soon with more detail and a larger time frame of data points.