Hiroshima and Nagasaki are not very good models to use when comparing to a BWR meltdown. The physics of the contamination dispersion are completely different. Much of the fallout from an atomic bomb is mixed with soil substrates and falls-out relatively close to the bomb impact - leading to a smaller contamination area.
In the case of BWR meltdowns, in contrast to an atomic warhead, a much larger portion of deposition takes place 10's,190's and even 1000's of miles from the accident as the contaminates are released as vapors and carried readily in atmospheric jet streams.
As for predicting the number of cancers from chernobyl, hiroshima, nagasaki etc, the real tragedy lies in the fact that we do not accurately record the actions of individuals following the event - because exposure boils down to an individuals personal behaviors. It would be like asking all the inhabitants of Northern Japan how much milk they have consumed since Fukushima failure - or more specifically how much milk was consumed, and where was the milk from, between march 11-13, 14,16, 23-24 etc. Without such detailed information we cannot accurately project any one individuals personal risk.