On the evening of April 14, 2026, severe thunderstorms barreled across southeastern Michigan, spinning up tornadoes that the National Weather Service had flagged as only a marginal risk less than a week earlier. By the time the storms cleared on April 15, the NWS Weather Forecast Office in Detroit/Pontiac was already bracing for what was coming next: a second, larger outbreak bearing down on the Upper Midwest just two days later. The back-to-back events tested a forecasting workforce that, by multiple accounts, had already shrunk by roughly 15% through a combination of voluntary buyouts and involuntary firings at NOAA, the NWS’s parent agency.
The result was a stress test that no one in the agency had asked for, arriving at the worst possible time.
Two outbreaks in four days
The first wave struck Michigan on April 14 and 15. Storm Prediction Center outlook graphics published by the Detroit/Pontiac office show how risk assessments shifted dramatically as the event approached. Six days out, the SPC placed the region under only marginal severe-weather risk. By Day 1, the outlook had escalated sharply, with tornado probabilities climbing as atmospheric ingredients came together faster than models had initially projected. Local forecasters were left compressing hours of warning refinement into a shrinking window as storm cells organized ahead of schedule.
Three days later, on April 17, a second and broader outbreak raked the Upper Midwest. The NWS Weather Forecast Office in La Crosse, Wisconsin, published a detailed event summary documenting tornado tracks, damage paths, and the meteorological timeline. That summary traces how forecast discussions moved from general severe-storm language to explicit tornado wording as confidence grew in the presence of strong wind shear and instability. The La Crosse record is the most complete federal account of what was forecast versus what actually happened on the ground.
Neither NWS event page has published final tornado counts with confirmed EF-scale ratings as of late May 2026, and official injury and fatality figures from both outbreaks have not appeared in a consolidated federal report. Those numbers matter: without them, the public cannot fully gauge the human cost or measure whether warning lead times were sufficient.
A workforce already running thin
The 15% figure describing NWS scientific staff losses has circulated widely in news coverage and congressional criticism of NOAA’s workforce policies, though no single federal document breaks down exactly how many employees departed through buyouts versus firings, or how those losses were distributed across the agency’s 122 forecast offices. Some offices may have lost senior forecasters or specialized severe-storm analysts while others remained closer to full strength. NOAA and the U.S. Department of Commerce, which oversees the agency, have not published office-level staffing data that would clarify the picture.
What is documented at the federal level is a broader pattern of dangerous understaffing. The U.S. Government Accountability Office published report GAO-25-108597, titled “Aviation Meteorologists: Urgent Actions Needed to Address Staffing Concerns,” which found that NWS meteorologist levels at aviation weather centers had fallen below thresholds set by an FAA-NWS staffing agreement. The GAO concluded that the shortfall creates operational and safety risks for time-sensitive forecasting. That report focused on aviation weather services rather than severe-storm operations, so it does not directly address tornado warning performance. But it confirms that a nonpartisan congressional watchdog considers NWS staffing levels a serious institutional problem, not merely a political talking point.
On Capitol Hill, Senator Patty Murray of Washington has been among the most vocal critics. Her Senate office released a statement in which fired NOAA employees described being dismissed with little explanation and later contacted about possible reinstatement. According to that account, NOAA reinstated some probationary employees it had already terminated twice, a cycle that underscores how chaotic personnel policy has become. Murray characterized the actions as “destructive mass layoffs.” That framing carries an advocacy purpose and should be understood as one side of an active policy fight, but the underlying facts about firings and partial reinstatements are confirmed by her office’s own public record.
The National Weather Service Employees Organization, the union representing NWS forecasters, has separately raised alarms about the impact of staff reductions on forecast operations, though detailed public statements tying specific staffing gaps to the April 2026 outbreaks had not been released as of late May 2026.
What the evidence does and does not show
The strongest pieces of evidence in this story fall into two categories: official NWS event documentation and the GAO’s independent audit. The La Crosse and Detroit/Pontiac event summaries are primary federal records created by the offices that issued warnings during the outbreaks. They contain radar imagery, meteorological data, and timeline reconstructions that allow anyone to trace how forecasts evolved, when confidence grew, and when language shifted from general severe-storm wording to explicit tornado threats. For readers trying to understand what forecasters knew and when they knew it, these documents are the most reliable starting point.
The GAO report sits in a different but equally credible tier. Its systematic data collection and interviews with agency officials give it institutional weight. But its scope is limited to aviation-focused centers, not the local forecast offices that issue tornado warnings. Readers should treat it as strong evidence that staffing problems are real and agency-wide, while recognizing it does not speak directly to what happened during the April outbreaks.
What is missing from the public record is just as important. No NWS internal memo, after-action report, or official agency statement has explained how the loss of scientific staff affected real-time forecasting during either April event. No federal document draws a direct causal line between specific personnel cuts and specific missed or delayed tornado warnings. The connection is plausible, supported by the GAO’s broader findings and by the observable fact that the SPC risk assessments escalated rapidly in the final hours before both outbreaks. But plausible is not proven.
There is also no public performance review comparing warning lead times or false-alarm ratios from the April outbreaks against historical baselines. Without that comparison, it is impossible to say with certainty whether the events represented a measurable decline in forecast skill or simply reflected the inherent difficulty of predicting tornadoes days in advance, especially in marginal early-spring atmospheric setups where small changes in moisture or wind shear can dramatically alter storm behavior.
Unanswered questions heading into peak tornado season
Tornado season in the central United States typically peaks between May and June, meaning the same forecast offices that scrambled through the April outbreaks are now entering the most demanding stretch of the year. If the 15% workforce reduction is as widespread as critics contend, the pressure on remaining forecasters will only intensify as severe-weather frequency climbs.
Several questions remain unanswered heading into the heart of the season. How many of the reinstated probationary employees have returned to operational forecasting roles, and at which offices? Has NOAA implemented any contingency staffing plans to cover offices that lost senior personnel? And will the agency release an after-action review of the April outbreaks that addresses whether staffing played a role in forecast performance?
Until those answers surface, the April 14 through 17 tornado outbreaks stand as the sharpest warning yet that a diminished NWS workforce may be approaching a breaking point. The storms did not wait for the agency to sort out its staffing crisis, and the next round will not wait either.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
