Fire moves fast.
Recovery doesn't.

A fire's damage gets counted twice — once while it's still burning, and again, much later, when the question changes.

In the days right after a fire, the priority is figuring out how bad it is and where. Much later, the question shifts from "how much burned" to "what's actually coming back."

The Palisades & Eaton case study belongs to the first kind of count. It's a from-scratch satellite burn-severity analysis — pulling Sentinel-2 imagery, computing a differenced Normalized Burn Ratio (dNBR), and mapping the fire scar without waiting on any official damage assessment. The question here is how close open data and open-source tools alone can get to the official numbers.

The Lahaina case study belongs to the second kind. Instead of the fire, it tracks individual building footprints — comparing what was still standing eleven days after the August 2023 fire against what was standing again in 2025. Two years is not long enough to finish rebuilding a town; drag the divider on that map and you can see exactly how far short of "recovered" it still is.

A note on the numbers: the building counts on this page are raw detections from a computer-vision model reading satellite imagery, not an official survey. The 848 → 952 figure is the net change in detected building footprints between the two dates — a reasonable proxy for rebuilding activity, not a precise count of which structures were rebuilt.