After your decade the radioactivity of a spent fuel assembly is 20,000 rem/hr @ 1 meter. You might have 200 of these in a big reactor--4 million rem/hr @ 1 meter if you could manage to get that close. We have an incident from Taiwan where people were living in buildings that emitted about .1 rem/day (although it decayed over time)--with the observed health effects over a decade being beneficial, not harmful. Thus I will call this a safe level. (Yes, I realize this sounds insane--all our data on the risks of radiation are based on acute exposure and this is just one of the pieces of data that suggests acute and chronic exposure have different effects.)
4 million rem/hr * 24 hr/day = ~100 million rem/day. Spread evenly over an area to produce this safe dose we have ~1 billion m^2 = a square 31 km on a side.
On timescale of months we have someone exposed to 10 rem/day with only minor health effects (he absorbed a dose that would be almost certainly lethal as a single exposure.) Spread evenly this is a square only 3.1 km on a side.
I'm not aware of any data that lets us set a low bound on a lethal zone. I would call 100 rem/hr certainly lethal--that's a mere 200m square.
In the real world you aren't going to get an even distribution within the death zone and nothing outside it--the actual death zone will be considerably smaller. I doubt it's going to extend much beyond the reactor complex itself.
As for the issue of the local water feature being contaminated--that's going to be dilute. The radioactivity might pose a cancer risk but it won't be a death zone and most of the radioactivity will be washed out to sea and diluted beyond detectability. This will make the death zone even smaller.