Your sources and the GP's don't actually conflict. They're just measuring different things.
The XKCD image is comparing Sieverts - absorbed dose - at a specific location. If you were staring into the core of Chernobyl, you received a massive dose. That effect is very localized.
The Scientific American article is comparing the regional effects: typical releases divided over a few hundred square miles.
The "coal far outweighs nuke" argument is based on global effects: While Chernobyl was intensely bad locally, the average effect over the whole surface of the Earth was very small - much smaller than the net emissions of coal.
These are all based on Sieverts/Rems. They measure instantaneous levels. The thing with coal is it's a gift that keeps on giving: the nuke shine of Chernobyl was intensely bad if you were there; a large number of becquerels (decays per second) of iodine-131 were released from Fukushima but it had an 8 day half-life. With coal it's a much longer half life and much more widespread. While there are no spatial and temporal hotspots like you have with nuke disasters, it keeps on going, so the everyone on the surface of the Earth will keep getting hit not just today, but for their entire lives.
The average human will absorb far more radiation from coal than from all nuke disasters. The numbers either way are small enough that you have other things to worry about as an individual, but coal is definitely going to cause a higher net number of cancers globally.