Brush Fires Don’t Have Addresses

Detection is the one piece of this job that got solved while we weren’t looking, and it got solved in our favor. NOAA’s Next-Generation Fire System can push an alert about a minute after a fire’s energy reaches the satellite, and ninety percent of the National Weather Service’s 122 Weather Forecast Offices have subscribed to the feed since it went live in February (NOAA, https://www.noaa.gov/news-release/noaa-unveils-powerful-convergence-of-ai-and-science-with-revolutionary-next-generation-fire-system). Camera networks are running detection across more than a thousand government cameras in the West, and those notifications routinely land five to twenty-five minutes ahead of the first 911 call (WFCA, https://wfca.com/preplan-articles/new-ai-technology-fire-service/). It’s federal, it’s free at the point of use, and departments with no budget line for any of it are already sitting inside the coverage.

I know it works because it worked on me last week.

It found me. It never found the call.

Two Minutes Was the Whole Margin

I got an ALERTWest notification on my phone for the butte in the middle of our town. It went out as a standard assignment, two engines, a brush unit, and a chief. I wasn’t the chief on that dispatch. But the notification came with a camera image and I could see the smoke in it, and I know what a fire on that ground does with those fuels, so I’d already decided that if it confirmed I was going. I got out the door while they were still tapping it out and added myself to the call.

Then the part that mattered. A caller had described the fire in the wrong place, and the location that went out with the assignment was wrong right along with it. Once I was outside and driving I could see smoke in the air, and I had the camera image on my phone to put it against. Between the two I could pin it. It was up near the columns, not where dispatch had it. I said so.

The Battalion Chief who caught the original dispatch was driving to the location he’d been given, the one the caller had described wrong, which took him around the backside of the butte. He told me afterward he wouldn’t have had eyes on that side for another few minutes.

I ended up first on scene and took command. Because we knew where the fire actually was, we got resources to the right side of it and kept it out of the trees.

That’s the whole story, and it’s smaller than it sounds. Call it two minutes.

Thirty Minutes Is a Rural Number

Almost everything written about early detection assumes a half hour and spends it on staging. Pre-position the brush units, warm up the mutual aid conversation, pull the preplans for the exposures downwind. All of that is real and all of it is a country problem. In a city district it’s fiction. A start on a butte inside city limits gets reported inside a couple of minutes because ten thousand people are looking right at it. You are never going to bank thirty minutes on that fire.

You’re going to bank two. Before last week I’d have told you two minutes isn’t worth writing doctrine around.

Two minutes is worth writing doctrine around. Just not for the reason the articles give.

What It Bought Was the Location

The value of that alert in a dense district wasn’t a head start on the clock. It was a picture I could trust.

Brush fires don’t really have addresses. They have locations, and that stays true inside city limits. What dispatch usually has is the caller’s address, which is where the person is standing, not where the fire is burning. There’s no clean way to translate what somebody is describing on the phone into a point on the ground, so what goes out with the assignment is an approximation and a direction. Then we do what we always do. We get into the area and try to see what the caller is seeing.

That step is invisible because it’s normal. Every one of us has run a block pattern through a neighborhood squinting for a column, working backward from a description a stranger gave a call taker. It isn’t a failure of dispatch and it isn’t a failure of the caller. Somebody looking at smoke from six blocks away is doing geometry under stress with no reference points. A camera on a fixed mount with a known bearing has none of those problems, and neither does a satellite.

So what I had on my phone wasn’t earlier information. It was truer information, and it let me skip the part where we go find out what the caller meant.

That distinction is the whole thing, because time and accuracy don’t compound the same way. Two minutes of head start compounds slowly and politely. A wrong location compounds fast. The first-arriving officer’s report shapes the next ten minutes of a wildland-urban call, and on a fuel-driven fire the gap between working the right flank and the wrong flank is the gap between a quick stop and a campaign. We spend enormous effort shaving seconds off turnout and travel. We have never had a lever on the accuracy of the location itself, and last week I watched one work.

The Picture Has No Seat on the Assignment

Here’s the uncomfortable part, and it isn’t the caller.

The good information was sitting on my phone. It wasn’t in the CAD. It wasn’t on the assignment. It wasn’t with the chief who actually caught the call. I had it because I subscribe to an alert, because I happened to be somewhere with a line of sight, and because I’m the kind of person who looks at his phone when it buzzes with a fire on it. The Battalion Chief driving the backside of the butte had none of those three things going for him, and he did absolutely nothing wrong.

That’s the gap worth naming. Detection reaches people, by geography and subscription and habit. It does not reach calls. NOAA’s feed goes to Weather Forecast Offices. Camera networks go to dashboards and to phones like mine. The work pushing past detection into predicted spread is headed for the same kinds of destinations (ESA EO4Society, https://eo4society.esa.int/2026/07/10/from-rapid-mapping-to-predictive-wildfire-intelligence/). None of it attaches to the call that gets created thirty seconds later. The moment that assignment goes out, the picture and the call are two separate objects moving through the world with no relationship to each other, and whether they ever meet depends on which chief happened to be holding the right phone.

The Cheapest Version of This Costs Nothing

I want to be straight about what I did, because it is not a model.

I jumped a call I wasn’t assigned to on the strength of a camera image and a column of smoke. It worked because I was right. If I’m wrong, I’m a Battalion Chief who self-dispatched, put a second command vehicle into a district that already had one coming, and stepped on another chief’s call for nothing. Nobody writes that into a procedure, and nobody should. What happened on the butte wasn’t a system. It was a guy with an app and twenty-four years of pattern recognition, and neither of those scales.

What does scale is deciding in advance that the picture belongs on the call.

That’s a doctrine decision well before it’s a software purchase. If a department subscribes to a camera network or sits inside a detection feed, the questions are small and answerable right now. Who is looking at that view when the tones drop. How does what they see reach the officer already driving toward a location somebody described secondhand. What is that person authorized to put on the radio without waiting to be asked. In most departments today the honest answers are nobody, it doesn’t, and nothing, and none of those answers cost money to change.

The real constraint isn’t technical and it isn’t budget. It’s that this sounds like extra work for somebody at the worst possible moment, and that acting on a picture you can’t confirm costs you something the day it turns out to be a controlled burn or nothing at all. That’s a leadership problem, and leadership problems are the ones we’re actually good at. We’ve done this before. We wrote the standard, made it the expectation, and stopped asking people to defend it call by call.

We Built Everything Around Not Knowing

For a hundred and fifty years this service has been organized around a single hard limit: we can’t know anything real about a fire until somebody tells us and we physically get there. Station placement, response standards, staffing models, run cards, all of it is an attempt to shrink a distance we could never actually reach into.

That limit broke quietly, and it broke in our favor. The satellites work. The cameras work. The alerts are going out right now, for free, over ground we already protect. The entire remaining problem is about thirty feet of routing between somebody’s phone and the call.

A picture that reaches a person is luck. A picture that reaches the call is doctrine.

We got lucky on the butte. I’d rather build it so we don’t have to be.

---

Robert Grand is a Battalion Chief at Eugene Springfield Fire with 24 years of service. He writes Frontline Intelligence, a newsletter on operational doctrine, technology, and leadership in Fire & EMS.

If this landed, share it with someone in Fire & EMS who needs to hear it.

#FireService #EMS #AI #FireTech #FireChief #PNW