On April 13 a strong tornado hit Ottawa, Kansas, but the public record from the Storm Prediction Center and local forecast offices shows the forecasting chain was more complicated than a single narrative blaming missing weather-balloon data.

At 7:23 p.m. on Monday, April 13, an EF-2 tornado formed a few miles west of Ottawa, Kansas, tracked east through town with peak winds of 125 miles per hour over a path of 7.3 miles, and was followed by a second EF-2 near Hillsdale and an EF-1 near Mound City. Three people were hurt in Ottawa. Nobody died.

By Thursday morning, NBC News ran a piece with the headline “‘We are missing data’: NWS weather balloon changes scrutinized as tornados hit Midwest,” linking staggered weather-balloon launches and staffing changes to a forecast that did not anticipate a Kansas City-area tornado threat. That argument traced the issue back to personnel decisions during the prior administration and included a mesonet meteorologist’s voice that, quoting him, said, “We are missing data at the normal times.”

The problem is that the forecast archive is public, time-stamped, and specifically addressed to the Kansas forecast offices. Reading the Storm Prediction Center outlooks and forecaster discussions for April 13 shows a chain of evolving forecasts and mesoscale notes that complicate the simple causal line the NBC story presented.

NBC’s story also leaned on the idea that many Great Plains offices shifted routine morning balloon launches to noon because of staffing, and that change explained gaps. A retired lab director quoted in the piece suggested continuing staffing shortages were driving the schedule change, and the piece linked that to buyouts and dismissals from the previous administration.

A spokesperson for the weather service, Erica Grow Cei, said the changing cadence of weather balloons has not affected forecasts, but the rest of the piece treated that statement as an afterthought. A quick check of the forecast archive would have shown the specific observational inputs and the real-time forecaster reasoning that day.

At 8:00 a.m. Central on April 13, the SPC issued the Day 1 Convective Outlook for 1300 UTC that placed the main Enhanced Risk over southern Minnesota and parts of Wisconsin. For the southern Plains the forecasters noted that whether sustained storms would even develop was “a key question,” and eastern Kansas did not appear in any tornado probability tier that morning.

An 11:36 a.m. update added a Slight Risk for parts of the southern Plains while hedging on specifics; the forecasters said confidence had “incrementally increased” but that “specifics regarding the extent of convective initiation are still uncertain.” Eastern Kansas remained outside the meaningful tornado probability area at that time.

A later update at 2:56 p.m. extended severe probabilities into eastern Kansas and said guidance and observational trends showed “an increased likelihood of a storm or two this afternoon.” That update arrived roughly four and a half hours before the Ottawa tornado, showing the risk area was changing through the afternoon.

Between 2:38 p.m. and 5:09 p.m. the SPC issued eight Mesoscale Discussions aimed at local offices. Seven of those discussions targeted other states where the environment looked far more dangerous, and MD #401 at 2:59 p.m. assigned an especially high watch probability to portions of Iowa and Minnesota, noting a strong tornado potential there.

MD #404, issued at 3:53 p.m., was the discussion that covered eastern Kansas and addressed the Springfield, Kansas City, Topeka, and Wichita forecast offices by their four-letter IDs. It listed a Watch probability of 20 percent and explicitly said, “Watch issuance appears unlikely at this time owing to uncertainty regarding the occurrence of convective initiation.”

Inside MD #404 the forecasters referenced a specific sounding: the 1 p.m. Central launch from Lamont, Oklahoma, and labeled it a “special sounding.” That term means an off-cycle launch ordered to capture extra data on a severe weather day, and Lamont is a research site that routinely launches four times a day independent of local NWS staffing.

The NWS upper-air network can support extra balloon releases “as needed” on top of the standard twice-daily schedule, and the closest NWS balloon sites to eastern Kansas include Topeka, Dodge City, and Norman. On April 13 SPC was reading sounding data from across the region, including launches the NBC piece suggested were missing.

MD #404 went out at 3:53 p.m. and the Ottawa tornado touched down at 7:23 p.m., so forecasters had roughly three and a half hours of real-time attention on the mesoscale setup, with afternoon sounding data already cited. That timeline ended in a 20 percent watch probability and a supercell that ultimately produced the tornado; it was a forecasting miss tied to the elusive trigger, not simply to an absence of balloon data.

That is the core challenge of mesoscale science: sometimes the atmosphere fails to break exactly where and when models and human judgment expect. The tools and observations were available in the record; the trigger location and strength outran current forecasting capability, and that is a technical limitation rather than a schedule argument.

Brad Temeyer, a meteorologist at the Kansas City NWS office, told NBC, “There was a pretty strong possibility there would be no showers or thunderstorms at all,” and he added, “It was a low-probability event of it occurring, but given that it did occur, it had high impact.” Those words underline how rare, high-impact misses can look in hindsight.

NBC also wrote, “Weather service offices did issue warnings when tornadoes were imminent, however.” The local forecasting offices in Topeka and Kansas City sent out several warnings later in the afternoon and evening, and those warnings helped get residents indoors and limit fatalities.

Other high-profile reporting runs have followed a similar script: pick a villain and build backward. After the July 4, 2025 Guadalupe River flood that killed at least 135 people, outlets framed staffing and leadership changes as the primary cause, but subsequent record evidence showed missed local responses and late evacuations also mattered. In that case, Camp Mystic did not begin evacuations until nearly two hours after local flash flood warnings, and the camp director testified he “did not see the early warnings.”

SPC’s April 13 archive shows the miss plainly: two forecasters signed the Kansas discussion — Chalmers and Hart — and their names sit with the record. The agency will post-mortem the case, verification will record the bust, and future forecasts will incorporate what April 13 taught the system.

The remaining question is whether newsrooms will treat the public record the same way institutions do: read the archive, report the miss, and hold their own work to the same standard they apply to others.