On a Monday night in April 2026, five tornadoes, including one particularly devastating touched down in the Kansas City area, wreaking havoc on buildings in Ottawa, Kansas.
While tornadoes are common in Kansas, the National Weather Service (NWS) seemed unprepared, issuing a forecast that afternoon that predicted no tornado activity.
The reasons behind such inaccurate predictions are complex.
Early weather balloon launches in the area experienced delays, potentially contributing to erroneous forecasts, which some blame on staffing shortages linked to budget cuts by the Trump administration affecting U.S. weather agencies.
In 2025, over 1,000 employees of the National Oceanic and Atmospheric Administration (NOAA), including many senior meteorologists at NWS, were laid off or accepted buyouts.
Since then, the administration has attempted to rectify this damage, scrambling to rehire hundreds of employees. Despite assurances from a NWS spokesperson that there is “no evidence of deterioration in NOAA’s weather model performance,” there remains concern.
However, independent meteorologists have voiced concerns about chronic understaffing at critical times following the budget cuts.
“Many who monitor severe weather closely find Storm Prediction Center forecasts less reliable than usual,” states William Gallus, a University of Iowa meteorologist.
This issue extends beyond minor inconveniences; accurate forecasts are essential to mitigate damage from extreme weather events, including intensifying hurricanes and record heat waves, both of which are becoming increasingly frequent due to climate change.
Moreover, the looming threat of a “Super El Niño” in the Pacific could lead to flooding along the West Coast and warmer global temperatures.
“Kansasans should never doubt the functionality of systems designed to protect them from severe weather,” declared Democratic Rep. Sharice David, representing the state affected by the tornadoes, in a statement.
This April, her office reached out to the Trump administration for information on the unexplained delays in weather balloon launches and their potential role in forecast errors. A month later, there has still been no response.
Clouds Gather
On paper, it may seem like weather agencies are on the mend. Congress largely disregarded requests for budget cuts, and an NWS spokesperson noted that they have hired 280 new employees since resuming hiring.
However, this hiring surge still leaves the agency with hundreds fewer employees than pre-cuts.
Even if the workforce is restored, replacing knowledgeable senior meteorologists will take time, according to Brian Tan, a meteorologist from the University at Albany, New York. “We’ve lost a wealth of organizational experience and expertise,” he says.
Craig McLean, former NOAA acting chief scientist, concurs: “This level of hiring highlights the detrimental impact of the Trump administration’s actions on the agency,” as reported by BBC Science Focus.
The missing staff collectively represent thousands of years of expertise in weather forecasting and climate modeling. “Losing 27,000 years of experience fundamentally changes the agency,” McLean states.
While experts do not anticipate these disruptions will cripple U.S. weather operations, they do believe the pace of improvements in forecast accuracy may decline. “We’re not heading into the dark ages,” Tan assures, pointing out that current models will likely continue to provide reliable predictions.
However, layoffs may hinder the speed at which forecasts become accurate, as this relies on research at NOAA labs and academic institutions, many of which experienced significant budget cuts last year, particularly at the Geophysical Fluid Dynamics Institute in Princeton, New Jersey.
Other researchers are concerned about plans to “dismantle” the National Center for Atmospheric Research in Colorado, a pivotal hub for climate and weather research; a consortium of over 100 universities currently manages the center. They plan to file a lawsuit to block this action by the Trump administration.
Into the Storm
Compounding uncertainties in weather forecasting is the emergence of AI. Several AI weather forecasting models introduced in recent years have shown superior performance compared to traditional models.
Traditional forecasting relies on supercomputers that simulate atmospheric behavior through complex equations. In contrast, AI models, developed from decades of historical data, learn to recognize patterns that predict specific weather conditions.
This shift promises greater efficiency, with some AI systems capable of operating on standard laptops rather than costly supercomputers. This advancement could enhance prediction accuracy and provide more localized information.
The Trump administration prioritized advancements at NOAA, enabling the integration of AI into predictive models. Nonetheless, officials clarify that this technology is intended to complement, not replace, existing forecasting tools.
NWS representatives note that the agency’s new AI model is “an addition to its suite of weather models, not a substitution,” as highlighted by BBC Science Focus.
Nevertheless, there’s growing apprehension about the potential for AI to dilute human input in forecasting, raising concerns about future report accuracy. Concerns persist regarding this trend.
“Humans play a crucial role, even just in terms of managing raw data,” asserts Jeffrey Schrader, a researcher studying weather forecasting at Columbia University.
His research indicates that forecasts derived from human meteorologists are typically 20% more accurate than those generated by statistical models. This discrepancy arises from meteorologists’ intimate understanding of local weather phenomena.
Forecasters familiar with their regions can, for instance, predict how local topography affects wind patterns and where rainfall may be underestimated by models. This nuanced understanding is something algorithms have yet to fully achieve.
“Experienced meteorologists bring immense value,” Schrader argues, noting that their role extends beyond mere interpretation of data and includes fostering relationships and trust within their communities, pivotal for community safety during severe weather.
Additionally, current AI models face limitations in predicting extreme weather, as they depend on datasets that may not encompass rare events. Research from a team of German and Swiss scientists found AI predictions underperformed relative to physics-based models when forecasting record-breaking conditions.
Notably, their accuracy declined as events intensified, highlighting the risks of relying solely on AI for predictions.
Inaccurate forecasts can have significant consequences, particularly as climate change exacerbates extreme weather events. Studies indicate that underestimating temperatures by even 1 degree can lead to increased mortality during heatwaves, with similar effects from underestimating cold waves.
“Without human oversight, AI predictions can become distorted and unreliable,” asserts Schrader.
Moreover, AI models cannot replace the necessity for basic weather observations, as even the most advanced AI systems require data from weather balloons and reconnaissance missions to function effectively.
“Technology is not a substitute for human involvement,” Tan emphasizes. “We need skilled professionals to interpret and act upon the data.”
Read more:
Source: www.sciencefocus.com
