Radiation risks

Or, how I stopped worrying and learnt to love the Fukushima Daiichi nuclear power plant.

Over the past rather abnormal week or so I've read a lot more about radiation and nuclear power than I ever thought I wanted to know. Mostly, I learnt that there are about half a dozen different and incompatible ways of measuring radiation, though the Sievert seems the most directly applicable, being a measure of the biological impact. Even this is sometimes presented misleadingly with people talking about Sv when they really mean a rate, Sv/h, and vice-versa. There's a typical example in this horribly confused Yomiuri article which talks about the likely fatal effects of 3,000mSv/h, when they presumably mean 3,000mSv total (over some short time). Another here in the Japan Times reports a reading of "0.000142 millisievert" in the Tokyo area, which obviously misses "per hour". It's possible that the Japanese media is particularly bad at this: their style is often to spew out context-free numbers, rather than actually educate (eg crime stats of foreigners, which are never actually directly compared to those of natives, because they are generally lower on a like-for-like comparison), and adding in mistakes makes it even harder to make sense of.

Xkcd (which I copied here), while not claiming to be a reliable resource, provides a convenient summary of a wide range of exposure factors and outcomes. David Spiegelhalter also has a good article here about the risks of Fukushima. He links to this article which suggests a 0.04% risk of cancer (29,000/70,000,000) from a CT scan, say 5mSv, or round it to 4 for convenience. That works out (under the linear assumption, which is disputed but let's gloss over that for now) to 1 cancer per 10Sv, whereas this perhaps somewhat rosy-hued article pegs the risk as 4% for 1Sv, or only 0.4 cancers per 10Sv. I'll stick with the worse figure and invent the new(?) unit of the microcancer (µC), analogous to micromort, and use 10µSv = 1µC. (The purpose of this post, as some of you may have guessed, is really just to show off that I've found that Option-m = µ on Macs.)

If we assume a typical cancer incidence of 25%, then a typical day of life averages about 9 µC (0.25 per 80 years). Typical background exposure is about 3mSv per year or 10 µSv per day, which would imply via my conversion factor that this background radiation is responsible for about 10% of all cancers. So far, these hand-waving approximations all seem to hang together surprisingly plausibly.

So, what about the effects of Fukushima? According to this (governmental) web site, 5 µSv/h is the the ambient radiation rate at which official measures have to be taken in Japan. That value is at the very top of the graph, with the observed data (for the prefecture I live in) nearly flat-lining at the bottom around 0.1µSv/h, though it's clearly been creeping up in the last couple of days. There were a couple of bigger bumps last week too which have dropped off the graph now, but nothing as high as 0.1µSv/h. There is certainly some leakage from the plant going on, that is reaching us. Other prefectures are a bit closer to the threshold (the parent page is here) and Fukushima, for which there is a pdf of lots of stations rather than a single graph, has much higher readings in some places.

5µSv/h is 120µSv/day or 43mSv/year, roughly the limit for workers at nuclear power plants. According to my conversion factor, this should translate to 0.4% additional cancer chance per year, which would add up substantially over a lifetime. Now, nuclear power station workers don't have such significantly elevated cancer levels, but they also probably don't actually get that much radiation either continuously throughout their entire careers - it's the limit, not a target. So I'm not ready to throw away the calculations yet.

100µSv/h sustained continuously with no attempt at mitigation would amount to about 70mSv in a month, and this would certainly mount up to a pretty substantial dose (and cancer risk) over time. Some of the Fukushima readings are up at that level around the edge of the 30km "stay indoors" region, and I don't think I'd want to stay there indefinitely if the level does not drop. However, even in this case, it would not amount to an immediately dangerous emergency requiring a panicked evacuation. It may be reasonable to stay there for a week or two, especially if it's possible to stay mainly indoors, and hope that the situation will be brought under control in that time. Assuming it is mostly iodine, the half-life is only 8 days so it won't hang around indefinitely.

It's also worth noting that cancer rates vary pretty widely from place to place and lifestyle to lifestyle anyway. Japan has famously high stomach cancer rates, for example, probably due substantially to the diet. Of course it has low rates for other types and a long life expectancy overall, but the point is that these risks vary substantially due to various lifestyle factors, that people show no signs of wanting to change. Some people even choose to smoke, probably because they are nervous about a nuclear catastrophe :-)


This page suggests about a 15% chance of lung cancer from smoking, so if we assume a 20 per day habit for 40 years, each cigarette is about half a µC, and the day's tally is 10µC or equivalent to 100µSv of radiation, a rate of 4µSv/h continuously through the day. So it seems that plenty of people find that sort of risk completely acceptable.

From the public health perspective, of course, these figures seem much more alarming. Even a single microcancer per person adds up to tens of cases over the Greater Tokyo area. And I think it's entirely right that there should be regulations to limit the extent to which TEPCO and similar can just dump pollutants on everyone. Beyond the health impacts, it is also (at least potentially) a major environmental problem for the immediate area, if substantial contamination does occur (or already has). Farming in the area may be completely wiped out. So I don't necessarily disagree with the Steve Bloom types who are going on about what a catastrophe it is. However, I'm approaching this analysis primarily from the selfishly personal level, as the British Embassy requested that I "consider leaving the area". I have done so (considered, that is, not actually left!), and it seems to me that the risks are entirely acceptable to me at the current level. By this I don't just mean the actual radiation which is currently negligible here, but the threat of an increase should things turn pear-shaped. I've got several orders of magnitude to play with before I need to feel worried. I can still resent TEPCO for imposing the risks on me, but reasonably decide that I should stay here rather than run away (taking a small additional radiation dose en route) for some uncertain, but potentially rather long, period of time.

If we assume (quite reasonably) that pedestrians suffer about 10 times as many serious injuries on the roads as deaths from traffic accidents, and (as a wild guess) that a typical pedestrian crosses a road every quarter-mile, then it seems from the micromort animations (1 micromort = 17 miles of walking) linked above that a single µSv of radiation exposure is quite literally not worth crossing the road to avoid. I'm equating a cancer with a KSI statistic here, which may be debatable but doesn't seem entirely ridiculous. Coincidentally, the peak radiation rate in Tokyo a few days ago, that had the media in a quite a tizzy, was a little less than 1µSv/h (and this level didn't last long). Hope no-one ran home in a panic.