Blog · November 2, 2026

Field notes: the forty minutes between the shake and the first usable pass

A remote satellite ground station with large dish antennas under a dark, moody sky.

The ground stops moving and the clock starts.

In the first ten minutes after a damaging quake, nobody is talking about imagery. Dispatch is pulling 911 logs, mutual aid is phoning in, and the EOC is counting heads. That's the right order of operations. But somewhere around minute fifteen, someone in the room asks the question that decides the shape of the next 72 hours: where do we send the first strike teams.

That's the forty-minute window I think about. Not the hour after, when ShakeMap grids and state-plane coordinates start filling the common operating picture. The window before that, when the only ground truth anyone has is phone calls, scanner chatter, and whatever a handful of first-in units can see from the cab of a truck. It's thin information, and it's directional at best.

A caller on the west side says a parking structure pancaked. A unit on the east side can't get a truck down a road because of debris. None of that tells you where the next collapse is, only where the first ones were loud enough to call in.

Event-driven tasking exists for exactly this gap. The moment a damaging event registers, VHR satellite assets get re-tasked off whatever they were doing and pointed at the affected footprint. That doesn't happen instantly, orbital mechanics and cloud cover don't care about your incident command structure. But the tasking request goes in during that forty-minute window, not after the first situation report gets typed up, because every minute the request waits is a minute added to when the pass actually happens.

What you get back, when the pass clears, is a structure-by-structure read of a neighborhood you can't drive through yet. Roofs that show collapse signatures. Blocks where the building envelopes look intact from above even if nobody's confirmed interiors on the ground. It won't tell you who's trapped. It will tell you which five-block radius has three collapsed structures and which has none, and that's the input that turns a search grid from a guess into a plan.

I've sat in EOCs where the team ran the first operational period on windshield surveys alone, because that's what was available, and watched teams get sent to a block that turned out fine while three streets over sat untouched until the second day. Nobody did anything wrong. The information just wasn't there yet. A dated structure-by-structure layer, even a rough first pass, changes that math. It lets a planning section lead say "concentrate here first" with something better than a hunch and a scanner feed.

Search and rescue staging areas get picked fast, usually based on road access and open lots, and teams push out from there toward wherever the calls are loudest. If the imagery shows damage clustered somewhere other than where the loudest calls came from, that's worth knowing before task forces commit hours to the wrong sector. The imagery doesn't replace a search dog or a hasty team on the ground. It tells the ops section which ground to send them to first.

Treat that window as a scheduled step in the activation plan, not a thing that gets remembered once the imagery becomes urgent. Put the trigger in the EOC's SOP: when a damaging event registers above a set threshold, the tasking request goes out within the first half hour, before the first situation report even circulates. Do that, and the imagery clears while task forces are still loading trucks at staging, not after they've already committed to a sector.

That's the gap Building Damage Classification is built to close, a dated, per-structure damage read delivered right after the event, so the first operational period runs on something better than a parking lot's worth of radio traffic.

If your activation plan doesn't yet include a standing tasking trigger for VHR imagery, that's worth fixing before the next one hits.