The only two Level 7 nuclear events in the world — Chernobyl and Fukushima Daiichi — have have rendered large swaths of land unusable for a significant amount of time. Whether or not the exclusion zones surrounding these two areas need to be as large as they are can be argued. However, when a hydroelectric dam fails more people die initially, but as the cleanup happens and the reporters fade away, the event fades into the background. With nuclear incidents people are persistently reminded whether through apocalyptic movies or people going into the exclusion zones to photograph the ruins that are left behind.
We could make movies about dams that collapse, too and we have. I'm not even sure there are more about nuclear accidents (nuclear war, yes, but accidents?). Even if there are, I think the movies are a symptom of the problem, not its cause. The invisible danger of radiation being held back by walls of concrete just looks more dangerous to us than the visible danger of a wall of water being held back by walls of concrete.
And Chernobyl? A nature reserve. I can see someone writing a SF story about cultures that protect existing nature by the controlled disposal of radio-active material.
1. It's out-of-the-ordinary. Nobody would pay attention to the monthly news report that coal burning has, yet again, damaged the environment to the same degree as a meltdown.
2. Our monkey brains are incredibly bad at realizing that gases can be more dangerous than those scary barrels with scary warning symbols. Instead of being hidden away underground, gasses conveniently "disappear" into the air.
There is no way to reliably estimate an event that rare. And to publish such an estimate at all is insanely irresponsible.
It's the same way one calculates the probability of being struck by lightening, for example.
The risk analysis is here: http://www.oecd-nea.org/ndd/reports/2010/nea6862-comparing-r...
In it, the probabilistic risk of a major nuclear accidents are calculated. It's a bit difficult to parse without a stats background, but if you Google the title there are some good summaries.
Most of the data is presented in graphs common in probabilistic risk analysis, but the authors do examine an example of the anyhow burg Nuclear Powerplant in Switzerland.
This is a very low baseline. The plant consists of Gen. II boiling water reactors, built in 1967. The reactor core has multiple cracks in the core shroud.
Even with this design, the authors describe how the risk analysis demonstrates that the probability of a major accident (deaths > 2000) is less than one per one million reactor years.
The risk graphs demonstrate that for reactors newer than Gen. II, the risk is greatly reduced, to the point that modern reactors are 1,600 safer than Muhlenburg.
The OECD estimates that the probability of an accident at a Gen. II or higher reactor that results in significant casualties is approximately one per one million years.
All of the accidents in the history of nuclear power have caused fewer deaths than the failure of a single hydroelectric dam. Combined these disasters resulted in an estimated 60,000 deaths.
That number pales in comparison to the 288,000 people who die every year from particulate inhalation due to fossil fuels.