Into the Storm: Advancing Tornado Prediction and Resilience

Each year, more than 1,200 tornadoes carve a path of destruction across the United States, endangering lives and devastating communities.

The U.S. National Science Foundation has long invested in groundbreaking research to better understand tornado behavior, improve storm tracking, extend warning times and design resilient infrastructure — safeguarding vulnerable communities and reducing economic losses.

Warning before the whirlwind

NSF has advanced tornado detection and forecasting for decades. Examples include:

A Doppler on Wheels mobile radar truck.

Credit: Ryan McGinnis

Doppler on Wheels

Since the 1960s, NSF has funded key advancements in Doppler radar technology, including the development of DOW in the 1990s. These truck-mounted mobile radars are designed to get as close to tornadoes as is safely possible. They allow researchers to observe tornadoes directly and collect detailed, real-time data on wind patterns, storm structure and dynamics.

NSF Center for Analysis and Prediction of Storms

CAPS, an NSF Science and Technology Center established in 1989, was among the first to demonstrate that it is possible to predict storm-scale weather. Its award-winning Advanced Regional Prediction System, designed for research and commercial use, revolutionized how meteorologists model and predict tornadoes. The center continues to refine and apply these systems today.

Engineering resilience

NSF supports advanced research facilities that develop and test building designs to protect communities better and reduce wind-induced damage from severe storms, including:

NSF Wind Hazard and Infrastructure Performance Center

WHIP, an NSF-funded Industry-University Cooperative Research Center, develops innovative solutions to enhance the wind resistance of buildings and infrastructure. Its research has improved design, testing protocols and materials that enhance the wind resistance of buildings, influencing national standards and building codes to make communities more storm resilient.

NSF Center for Collaborative Adaptive Sensing of the Atmosphere

Established in 2003 as an NSF Engineering Research Center, CASA developed networks of short-range radars to monitor the lower atmosphere with rapid, high-resolution data on storms, winds and precipitation. The center's innovations advanced severe weather detection and forecasting, bolstering public safety and informing building design practices.

NSF Wall of Wind

The NSF Natural Hazards Engineering Research Infrastructure (NSF NHERI) Wall of Wind Experimental Facility can generate winds exceeding 157 mph, allowing researchers to test full-scale buildings and develop impact-resistant roofing systems, stronger roof-to-wall connections and debris-resistant building envelopes.

NSF Boundary Layer Wind Tunnel

This 40-meter wind tunnel, supported by the NSF NHERI program, simulates turbulent wind conditions like those in tornadoes and hurricanes. Researchers study how wind interacts with various building shapes, helping engineers improve designs to withstand extreme winds better and reduce storm damage.

Twelve giant yellow fans inside a large warehouse.

The Florida International University Wall of Wind facility.

Credit: Florida International University

Improving tornado science through models, data and AI

From advanced radar tools to artificial intelligence, NSF-supported research has transformed how scientists understand, detect and respond to tornadoes.

Computer simulation with orange and reddish streaks.

A virtual tornado created using some of the most powerful supercomputers in the world.

Credit: Leigh Orf / University of Wisconsin-Madison

NSF-supported scientists created a polarimetric radar simulator to better understand how debris from tornadoes appears in radar data. By analyzing how different debris can scatter signals, the research helped explain complex radar signatures seen during storms. These advances improved tornado detection and supported quicker, more accurate assessments of damage in affected communities.

In 2020, NSF-supported researchers created one of the most detailed simulations of a tornado at the time, capturing its structure and evolution at unprecedented resolution. These simulations helped scientists better understand how tornadoes form and behave, improving forecasting models and severe weather prediction.

Beginning in 2022, the Propagation, Evolution, and Rotation in Linear Storms field campaign is improving the ability to predict dangerous tornadoes that form within fast-moving storm lines — events that are often difficult to detect and can strike with little warning. By collecting detailed observations and linking them to real-world impacts, researchers are refining forecasting models and strengthening early warning systems. These advances are helping provide communities with more accurate, timely information to reduce injuries and save lives.

In 2025, NSF-supported researchers developed an AI-driven approach to rapidly assess tornado damage and forecast recovery using satellite imagery and deep learning. Trained on post-disaster images, the model classified building damage and estimated recovery timelines, helping communities to respond and rebuild faster.

Today, the NSF AI2ES Artificial Intelligence (AI) for Earth system (ES) is integrating advanced machine learning approaches into severe storm forecasting, enabling faster data analysis and more accurate, timely tornado warnings. These continuous investments — spanning mobile radars, advanced weather models and cutting-edge AI — are improving forecasts, extending tornado warning lead times and ultimately helping to save lives.