Pull archived NEXRAD Level-II sweeps around a candidate fall's time and location.
Finding meteorites in weather radar
MetDetect: automatically detecting freshly fallen meteorites in the U.S. NEXRAD Doppler weather-radar network.
When a fireball drops meteorites, the surviving stones fall through the atmosphere as a dispersed cloud of debris. As they pass through the beam of a nearby weather radar they briefly light up the lowest scan elevations, a faint, falling signature buried in the same data meteorologists use to track storms.
The U.S. NEXRAD network archives every sweep from 150+ radars, going back decades. Searching it by hand is how several meteorite falls have already been recovered, but it's slow and easy to miss. MetDetect automates that search: a machine-learning pipeline that scans the archive around candidate fall times and flags the radar signature of falling meteorites, so we can point recovery teams to a fall while the meteorites are still fresh.
How it works
Isolate low-altitude reflectivity gates consistent with falling debris, rejecting weather and clutter.
Score and cluster the surviving gates to localize the fall and estimate a search footprint.
Flag the strongest candidates for human review and field recovery.