That is my understanding as well--the actual source states "affected" instead of "died." I have yet to see anyone actually give a reliable estimation of how to turn "affected" into "died", and I can sustain arguments for anything between 1% and 100%.

There are essentially three pieces of information going into this estimate: pre-fire density, fire acreage, and fire mortality (in the long run). The first number is known at best to an order of magnitude, the second is quite well-known (thanks to satellites), and the last appears to be unknown.