On the evening of April 25, 2025, a cluster of supercell thunderstorms ripped across parts of the southern Plains, producing tornadoes that struck communities with little advance notice. It was the kind of outbreak that the National Weather Service exists to anticipate. But at several NWS forecast offices in the region, the ranks of experienced meteorologists and technicians had been thinned by months of federally encouraged buyouts, and at least some stations had already scaled back the atmospheric observations that feed the computer models used to predict exactly this kind of event.
By mid-2026, the consequences of that workforce drain are becoming harder to ignore. The NWS Employees Organization, the union representing the agency’s frontline staff, has warned that roughly 15 percent of the agency’s scientific workforce has departed through buyouts, early retirements, and attrition since the current round of federal downsizing began. That figure is a union estimate and has not been confirmed by an official NOAA personnel audit or inspector general review. Regardless of the precise number, dozens of the 122 local forecast offices are operating below minimum staffing levels, and the operational fallout is landing squarely on the data collection that underpins tornado forecasting.
Weather balloons are going unlaunched
Twice a day, at coordinated times around the globe, NWS technicians release helium-filled weather balloons carrying instruments called radiosondes. Each radiosonde measures temperature, humidity, pressure, and wind speed from the ground up through the stratosphere, building a vertical snapshot of the atmosphere that no satellite or ground radar can fully replicate. These profiles are direct inputs to the numerical weather prediction models that generate severe thunderstorm and tornado forecasts.
When a local office crosses a certain vacancy threshold, internal NWS directives require it to cut balloon launches from twice daily to once daily. At even higher vacancy levels, launches are suspended entirely. These are not hypothetical contingencies. Multiple offices have already triggered the cutbacks, some during active severe weather periods when the data gaps matter most.
A December 2025 analysis published through the University of Georgia’s Cooperative Extension climate program confirmed that the staffing losses were still accumulating at that point, with some stations unable to launch balloons at all for lack of trained personnel. The analysis, written for agricultural producers and emergency managers who depend on accurate forecasts for planting decisions and storm preparedness, described the situation as ongoing and worsening rather than a one-time disruption.
The data gap hits tornado forecasting hardest
Tornadoes are among the most difficult weather phenomena to predict because they depend on fine-scale atmospheric conditions: the exact profile of wind shear, the depth of moisture, the height of the cap that suppresses or releases convective energy. Forecasters assess these variables using a combination of surface observations, radar, satellite imagery, and upper-air soundings. Remove the soundings, and the models that stitch everything together lose a critical anchor point.
The mechanism is well understood among atmospheric scientists. Fewer upper-air observations degrade model initialization, which can shorten warning lead times and increase the probability of a missed or late tornado alert. In marginal setups, where a small shift in wind shear or instability determines whether storms stay disorganized or become violently tornadic, a single missing radiosonde profile can be the difference between a confident watch issued hours in advance and a warning fired off minutes before a funnel touches down.
No published after-action report has yet quantified exactly how many minutes of lead time were lost in a specific 2025 or 2026 outbreak because a nearby station skipped its launch. That level of granular verification would require internal NWS performance data that the agency has not released. But the causal chain is not speculative: the NWS’s own technical documentation describes radiosonde data as essential to the models it uses for severe storm prediction. The agency is now collecting less of the data it says its own systems need.
Uneven pain across the map
The staffing losses are not distributed evenly. Offices near large metropolitan areas or university research hubs have generally retained more experienced staff, while smaller offices covering vast rural territories have lost multiple senior forecasters and technicians in quick succession. Without a detailed, office-by-office staffing map, which the NWS has not made public, it is difficult for outside observers to know precisely which communities face the steepest risk from degraded services.
That geographic unevenness matters because rural communities in the Great Plains and the Deep South are often the ones most exposed to violent tornadoes and least served by redundant warning infrastructure. A town that depends on a single NWS office for its watches and warnings has no fallback if that office is running a skeleton crew during a major outbreak.
The budget picture adds further uncertainty. The Congressional Research Service has published a high-level overview of NOAA’s fiscal year 2026 budget request, which proposes significant changes to the agency’s spending across satellite programs, research, and operations. But the document does not specify how many forecaster or technician positions would be added or cut at local offices. Whether Congress will approve, modify, or reject the request remains an open question as of June 2026, and the final appropriation could look very different from what was proposed.
What residents and local officials can do now
For the millions of people living in tornado-prone regions, the practical reality is this: the warning system they depend on is operating with fewer scientists, less atmospheric data, and thinner margins for error than it was just a few years ago. That does not mean every storm will be missed. Many NWS forecasters continue to work grueling shifts monitoring radar, interpreting models, and issuing warnings when storms develop. Their skill and dedication have not changed.
But residents should not assume the system will always deliver the same advance notice they have come to expect. Emergency managers in several states have already begun adjusting their planning, placing greater emphasis on redundant information channels: trained local spotter networks, private-sector radar analysis, and regional coordination among neighboring forecast offices that can share workloads during outbreaks.
Public education may be the most immediate lever available. When warning lead times shrink, individual response time becomes the variable that saves lives. Knowing where to shelter, having multiple ways to receive alerts, and understanding that a watch can escalate to a warning in minutes rather than hours are all pieces of preparedness that do not depend on federal staffing levels.
A safety net stretched thinner than the public realizes
The staffing crisis at the National Weather Service is not an abstract budget dispute. It is a measurable reduction in the observations that feed the models that produce the warnings that give people time to get underground. The agency built formal triggers for service cutbacks into its own operational playbook, and those triggers have now been crossed at multiple offices during severe weather season. Until the workforce is rebuilt and full upper-air coverage is restored, communities across the central United States are living with a warning system that is more fragile than at any point in recent memory. The margin between a timely alert and a deadly surprise has gotten narrower, and the people who stand to pay the price are the ones least likely to know it.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
