Different definitions and means of reporting by PR agencies and officials.
The biggest is the difference in "caused by" and "connected with". The first is a fairly small number. The second is much larger, and most often has to do with damage to infrastructure, things like no electricity. Even then, the number we hear cited aren't real. They're often the numbers cited in the studies, but the studies always give ranges because they're based on statistical analyses and that means probabilities. We hear things like, "The number is 900 to 3500, with 2700 being the most likely number", but "most likely" doesn't necessarily mean even "50% confident." Sometimes we hear the most likely number cited, sometimes the extreme that fits the argument; if we hear the range, we remember the number that fits our bias.
We're of two minds when it comes to such things: When the feds ride roughshod over locals, esp. (D) locals, and officials "of color", we get pissed off because the assumption the locals can't manage better is offensive, if not racist; but when the feds leave locals to implement their own plans, as the law and regs require, and the locals screw it up, we get pissed off that the feds didn't take over.