This interview takes place in April 2030. It hasn't happened yet.

Jolene Rader's office in the Sedgwick County Emergency Operations Center has four screens mounted behind her desk. Two show radar feeds. One shows a county map gridded into response zones. The fourth used to display Warn-on-Forecast probability fields, experimental data from NOAA's National Severe Storms Laboratory that could project tornado threats up to an hour out.¹ It now shows the local news.

Rader has been the county's Emergency Management Director since 2019. She came up through fire services, drinks coffee that has clearly been reheated more than once, and checks her phone at intervals that suggest a body trained to expect interruption. A framed SKYWARN Recognition Day certificate from 2023 hangs near the door. There is no certificate for 2024 or after.

She agreed to talk between what she calls "the quiet hours," the gap between morning briefings and the afternoon convective ramp-up, when her phone starts buzzing and doesn't stop.

When you started this job, what was the number you built everything around?

Jolene: Thirteen minutes.² National average lead time on a tornado warning. You plan around it, you drill around it, you build your whole response architecture around it. Thirteen minutes means I can activate sirens, push alerts, get spotters on the phone, and still have time for people to get to shelter. Not comfortable time. But enough.

NWS guidance actually told us to plan on five to ten, with contingencies for less.³ But thirteen was the average, and you could work with thirteen.

What's the number now?

Jolene: I don't have a number.

That's part of it. The research infrastructure that would tell me what the new average is got cut too.⁴ NSSL is gone. The people who tracked warning performance, verification stats, all of that. So I'm working off feel. And feel says I've lost minutes. How many, I can't give you a sourced figure. I can tell you that last May I had a warning come in and by the time I confirmed it with my spotter network, the storm was already past the first response zone.

That didn't use to happen.

Walk me through what happens now when a tornado warning drops for your county.

Jolene: OK. So. A warning comes in from the local forecast office. And I should say, the people still working there are good. They're working harder than they've ever worked. But there are fewer of them, and the data they're working with has holes.⁵ The balloon launches that used to fill in the atmospheric picture between radar scans, a lot of those are gone.⁶ The electronics techs who kept the radar and weather radio transmitters maintained, reduced.⁷ So the warning arrives, and it's less confident than it used to be. You can feel it in the language. More hedging. Wider polygons.

Warning drops. I've got my polygon on screen. First thing: where is it relative to population density. Sedgwick County has almost half a million people, but they're not distributed evenly. Wichita's dense. The rural parts aren't. Mobile home parks are my highest-vulnerability targets. Those folks die at wildly disproportionate rates and they have the least shelter access.

So I'm making calls simultaneously. Sirens go county-wide within the polygon, that's automatic now, we took the manual trigger out of the loop in '27 because I couldn't afford the thirty seconds. Wireless Emergency Alerts push to phones, but that message is generic — "Tornado Warning in this area until 4:45 PM" — and I have zero control over it.⁸ My value-add is the targeted stuff. Calling mobile home park managers directly. Getting my SKYWARN spotters on the ground to confirm rotation. Feeding real-time ground truth back to the forecast office because they need my eyes now more than they used to.

And here's where it gets — I don't want to say ugly, but.

When I had thirteen minutes, I could do all of that. When I have six or seven, I'm triaging. The mobile home park on South Hydraulic gets my first call. The subdivision in Bel Aire with basements gets my attention second. I'm making that calculation based on vulnerability and shelter access, and it's the right call, but nobody elected me to decide whose warning comes with a phone call and whose comes from a phone buzz.

More than half of tornado warnings have historically been false alarms.⁹ Has that ratio changed?

Jolene: OK, so, false alarms aren't a bug. They're a feature of a system trying to err on the side of not killing you. You warn wide, you accept that some won't verify, because the alternative is missing the one that does.

But when your upstream data degrades, you get two bad options. One: warn on less confidence, more false alarms, people stop sheltering because they've been burned six times this season. Two: wait for higher confidence, shorter lead time, the warning's accurate but nobody has time to act on it.

Pick your poison. Both kill people. Just different people, different ways.

I've watched compliance drop. Anecdotally, because again, nobody's formally tracking this anymore. But I talk to my spotters, I talk to school administrators. People are slower to move. They check their phones, they look out the window, they wait. And I get it. If you've been told to shelter six times in April and nothing happened, the seventh time you're going to pour yourself a coffee and glance at the sky. The seventh time might be the EF3.

What do you tell city council when they ask why the system failed?

Jolene: [long pause]

I show them the org chart. I show them what the NWS office looked like in 2023 and what it looks like now. I show them the maintenance schedule for the WSR-88D and the gaps where the electronics tech position hasn't been filled. I'm very boring about it. Spreadsheets. Lots of spreadsheets.

Last year a councilmember asked me if we could "supplement with private-sector weather data." I said sure, absolutely, if the county wants to pay for a subscription that costs more than my entire department budget, I'd love to have that conversation. Let's put it on the agenda.

That got quiet fast.

You were supposed to have Warn-on-Forecast by now. Sixty-minute lead times.¹⁰

Jolene: Yeah.

What keeps you in this job?

Jolene: My spotters. I've still got a SKYWARN network of about forty people in this county. Volunteers. Retired firefighters, ham radio guys, one man who I swear just likes standing in fields during severe weather and found a socially acceptable excuse. They go out in the worst conditions on earth because I ask them to, and they call in what they see, and that ground truth is sometimes the only real-time data I have that I fully trust.

They're my human sensor layer. They work for free.

When you ask what keeps me here. It's that I have forty people who will drive into a supercell because their neighbor needs to know it's coming. The day I don't have that, I'll go. But I'm not going to be the one who leaves while they're still out there.

Footnotes

1. NOAA's National Severe Storms Laboratory developed the Warn-on-Forecast program to extend tornado, severe thunderstorm, and flash flood warning lead times up to an hour. The program was projected for elimination under proposed NOAA budget cuts. https://www.nssl.noaa.gov/projects/wof/ ↩

2. The national average lead time for tornado warnings is 13 minutes. https://www.noaa.gov/stories/tornadoes-101 ↩

3. NWS operational guidance advises emergency managers to plan on 5–10 minutes of advance warning, with contingencies for less. https://www.weather.gov/lwx/IdentifyThreatsTOR ↩

4. The National Severe Storms Laboratory, which tracked warning performance and developed next-generation forecasting, was slated for elimination under proposed NOAA budget restructuring. https://www.wglt.org/local-news/2025-07-14/potential-noaa-weather-research-cuts-could-have-consequences ↩

5. Approximately 550 NWS employees departed through deferred resignation or were fired, leaving the agency down more than 10% of staffing heading into severe weather season. https://www.firstalert4.com/2025/05/28/5-former-nws-directors-call-protection-national-treasure-after-proposed-cuts/ ↩

6. NWS reduction in weather balloon launches left forecasters with significant data gaps during tornado outbreaks. https://www.nbcnews.com/science/science-news/meteorologists-say-nws-cuts-degraded-forecasts-recent-storms-rcna202386 ↩

7. NWS electronics technicians maintain radar and weather radio transmitters; their reduction was identified as a direct threat to warning system reliability. https://www.politifact.com/article/2025/mar/26/noaa-firings-came-before-us-tornado-outbreak/ ↩

8. Wireless Emergency Alerts push a generic tornado warning message to cell phones within a warning polygon. Local emergency managers have no control over message content. https://www.weather.gov/lwx/IdentifyThreatsTOR ↩

9. More than half of NWS tornado warnings are false alarms. https://www.weather.gov/bmx/research_falsealarms ↩

10. Warn-on-Forecast was designed to help forecasters issue hazardous weather warnings up to an hour in advance, with documented cases of emergency managers using early data to sound sirens before formal warnings, credited with saving lives. https://www.nssl.noaa.gov/projects/wof/ ↩