What Do Solar Storms Mean for GPS and Self-Driving Cars?

Key Takeaways

• A November 2025 geomagnetic storm pushed GNSS position errors above 10 meters across parts of the United States.
• The study shows a navigation hazard for autonomous transport, but it does not document a self-driving crash.
• Safer autonomy depends on sensor fusion, integrity checks, space-weather awareness, and independent PNT backups.

What Did the November 2025 Storm Actually Do to GPS?

On August 29, 2026, researchers from The Aerospace Corporation, Boston University, Boston College, and MIT published a Geophysical Research Letters study showing that the November 2025 geomagnetic superstorm produced Global Navigation Satellite System (GNSS) positioning errors exceeding 10 meters in parts of the continental United States. The result makes the connection between GPS and self-driving cars more concrete because the measured error was large enough to move a computed vehicle position far beyond the accuracy expected for lane-level automation. The research team tied the degradation to auroral activity and irregular structures in the ionosphere rather than to a malfunction of the GPS satellites themselves.

The storm developed after a sequence of coronal mass ejections reached Earth. NOAA had issued a G4 severe storm watch for November 12, 2025, and G4 conditions were recorded at 01:20 UTC that day. NOAA later reported that the storm peaked on November 12 and produced auroras visible as far south as Florida and Texas. The new research focused on what happened to radio-frequency navigation transmissions as the disturbed ionosphere spread much farther south than the regions where strong scintillation is more commonly expected.

The study combined GNSS observations with optical measurements of auroral brightness and other ionospheric data. Researchers found an east-west band of intense auroral activity stretching across much of the country. That activity created sharp electron-density gradients and smaller irregularities in the ionosphere. Transmissions from GPS and Galileo satellites passed through those structures, producing strong fluctuations in radio amplitude and phase. The paper reports S4 amplitude-scintillation values above 0.5 and horizontal positioning errors that exceeded 10 meters in affected regions for hours.

The finding deserves careful interpretation. It shows that severe space weather can corrupt a high-precision navigation solution over a broad mid-latitude region. It does not show that every receiver in the United States was wrong by 10 meters at the same instant, and it does not show that an autonomous vehicle crashed because of the storm. The research establishes a physical pathway from geomagnetic activity to navigation degradation, then identifies autonomous transportation and precision agriculture as sectors exposed to that pathway.

How Can a Solar Storm Move a GPS Position by More Than 10 Meters?

GPS determines position by measuring the travel time of radio transmissions from satellites in medium Earth orbit. A receiver combines broadcasts from multiple satellites whose positions and timing are known with great precision. The calculation assumes that the radio path can be modeled well enough to correct for predictable delays. Under ordinary conditions, modern receivers can compensate for much of the ionosphere’s effect, and higher-precision systems can use multiple frequencies, reference stations, correction services, carrier-phase measurements, and inertial data to achieve far better accuracy than an ordinary consumer device.

A geomagnetic storm changes that radio path. Energy from the solar wind enters Earth’s magnetic environment and can drive large changes in the ionosphere, the electrically charged region of the upper atmosphere through which GNSS transmissions travel. NOAA describes ionospheric scintillation as rapid modification of radio waves caused by small-scale electron-density structures. Severe scintillation can prevent a GPS receiver from maintaining lock on a transmission. Less severe scintillation can reduce position accuracy or the receiver’s confidence in the computed solution.

The November 2025 event was unusual because strong amplitude scintillation spread through mid-latitude regions across a broad span of the United States. The auroral oval shifted toward lower latitudes, producing intense auroral precipitation and a long band of steep ionospheric density gradients. Smaller structures formed within that disturbed region. When GNSS radio waves crossed those structures, diffraction and refraction changed the radio amplitude, phase, and apparent path length. Carrier tracking became less dependable, and position calculations degraded.

This distinction matters because the satellites can remain healthy even when users on Earth receive bad navigation information. GPS performance is commonly discussed as a property of the satellite constellation, but a user position also depends on the atmosphere, satellite geometry, local obstruction, multipath reflections, receiver design, correction data, and radio-processing software. The U.S. government reported that 2024 GPS service performance met a global-average 95% horizontal position-service accuracy metric of 8 meters under normal operating conditions. Severe storm-time ionospheric errors belong to a different operating regime, one in which propagation conditions can dominate the position solution.

New Space Economy coverage of solar storms and the space economy has described the same dependence from an industry perspective: satellite navigation is an Earth service whose reliability can be degraded without any navigation spacecraft being physically damaged. That distinction is important for risk planning because replacing satellites would not solve an ionospheric propagation event. The response has to include better monitoring, better receivers, alternative positioning inputs, and operational procedures for periods when satellite-derived location becomes less trustworthy.

What Does the Study Mean for GPS and Self-Driving Cars?

A 10-meter position error is large compared with the geometry of a road. Federal Highway Administration material identifies 3.6 meters, or 12 feet, as a common lane width on higher-speed roads. A GNSS solution displaced by more than 10 meters can place a computed position in a neighboring lane, on a shoulder, on a frontage road, or outside the mapped road corridor altogether. That scale explains why the Geophysical Research Letters authors describe the observed errors as sufficient to disrupt autonomous transportation.

The headline risk needs a second layer of analysis. Production autonomous driving systems do not normally steer by treating raw GPS coordinates as a complete description of the road. High-automation systems combine multiple information sources, and their localization software tries to determine both absolute geographic position and position relative to lanes, curbs, signs, buildings, and other mapped features. Waymo, for example, states that its Driver does not rely solely on GPS. It matches detailed maps with real-time data from lidar, cameras, radar, and onboard computing to determine road position.

That architecture changes the failure scenario. A bad GNSS fix does not automatically command a vehicle to jump 10 meters sideways. A well-designed system can compare satellite navigation against inertial sensors, wheel motion, lidar localization, visual features, map geometry, radar observations, and other data. If one source becomes inconsistent with the others, software can reject it, reduce its weighting, declare degraded localization, slow down, pull over, or stop service within the operating rules designed by the developer. Waymo’s current description of its technology emphasizes complementary sensor inputs and redundancy rather than dependence on one source.

The danger appears when several conditions align. GNSS can become misleading rather than simply unavailable. Other localization sources can also be degraded by tunnels, heavy precipitation, dirty sensors, snow-covered lane markings, repetitive road geometry, construction, map changes, or urban obstruction. A system that gives too much confidence to a corrupted navigation source can develop a confident but wrong estimate of its location. The U.S. Department of Transportation PNT plan identifies loss of an accurate position, velocity, and time solution as a public-safety concern for highly automated transportation systems, including cases involving misleading GNSS information.

The November research strengthens the case for treating space weather as part of the localization fault model. Vehicle developers already test for blocked satellites, urban multipath, jamming, spoofing, sensor failure, and map mismatch. Severe ionospheric scintillation adds another source of correlated, regional degradation. It can affect many vehicles over a large area at roughly the same time, which makes it different from a broken antenna or a single bad sensor on one vehicle.

Why Sensor Fusion Changes the Safety Interpretation

Autonomous driving depends on localization, perception, prediction, and planning working together. GNSS contributes geographic position and timing, but the driving task also requires a vehicle to understand lane boundaries, traffic lights, pedestrians, road edges, other vehicles, temporary construction, and the motion of objects nearby. Those functions are supplied by a combination of sensors and software rather than by satellite navigation alone.

A 2026 autonomous-driving sensor survey describes production stacks that combine cameras, lidar, automotive radar, and GNSS with inertial measurement units. Each sensing method has different failure modes. Cameras provide rich visual information but can lose performance in glare, darkness, fog, or occlusion. Lidar supplies precise three-dimensional geometry but can be affected by weather, contamination, or sparse returns from some surfaces. Radar measures range and velocity and can perform well in conditions that reduce camera visibility, yet its spatial interpretation differs from lidar and vision. Inertial sensors can bridge short GNSS interruptions, but their position error grows over time unless corrected.

Localization engineering consequently revolves around uncertainty as much as accuracy. A position estimate should include confidence information and integrity checks that tell the system whether the estimate can be trusted. The difference between an obvious outage and a plausible but wrong solution is important. Loss of satellite lock can trigger a straightforward fallback. A degraded solution that still looks valid can be harder to detect, so cross-checking independent sensors becomes a safety function rather than a performance enhancement.

Research predating the November 2025 storm already showed that tightly coupled inertial navigation and GNSS can lose required availability in dense urban settings. Work presented to the Institute of Navigation in 2021 evaluated self-driving localization in downtown Chicago using GNSS, inertial navigation, wheel-speed sensors, and vehicle constraints. The researchers found that vehicle speed and the duration of GPS-denied areas strongly affected navigation performance, and they examined where external ranging would be needed to maintain continuous integrity.

Space weather adds a regional atmospheric cause to that engineering problem. Urban blockage tends to be associated with specific streets, structures, tunnels, and local geometry. Geomagnetic disturbances can affect large areas at once and can alter radio quality even under an open sky. That means an autonomous system that passed localization testing in clear-sky rural conditions still needs a strategy for days when the ionosphere itself becomes the source of error.

For developers and regulators, the relevant metric is not whether GPS is “working.” The more useful questions are whether the vehicle can detect degraded navigation, how quickly it can isolate the faulty source, what independent sensors remain available, what level of localization uncertainty is acceptable for the current maneuver, and what minimum-risk behavior begins when confidence falls below that limit.

Why the May 2024 Farm Disruption Matters to Autonomous Transport

The November 2025 study was not an isolated warning. Research published in 2025 on the May 10-11, 2024 Gannon geomagnetic storm found precise-point-positioning errors reaching 70 meters in the central United States. The same May 2024 GPS study reported 10-20 meter errors in the Southwest and position errors around 10 meters during auroral activity on May 11. Researchers linked the worst central U.S. errors to steep ionospheric plasma gradients, strong scintillation, carrier-phase cycle slips, and unstable precise positioning.

Agriculture supplied a real operational demonstration because modern planting equipment can depend on centimeter-level GNSS guidance. During the May 2024 storm, farmers reported automated guidance failures during planting. NOAA later stated that the event caused more than $500 million in potential profit losses for U.S. farmers. The scientific GPS study establishes the magnitude and physical cause of the positioning degradation. The economic number comes from agricultural analysis and should be treated separately from the navigation measurements.

This comparison matters for autonomous road transport because precision agriculture and automated vehicles share part of the same dependency. Both can require continuous, trustworthy position estimates rather than occasional map guidance. Both may use high-precision corrections that rely on carrier-phase tracking. Both can suffer when the ionosphere introduces fast-changing errors that exceed the assumptions built into normal correction models. A tractor has the advantage of operating in a relatively controlled field with lower interaction density than an urban road. A road vehicle must manage moving traffic, pedestrians, junctions, and map constraints at the same time.

The May 2024 event also shows that timing matters economically. A navigation interruption during a low-demand period may produce limited losses. The same technical failure during planting, an airport arrival surge, a port bottleneck, or a period of heavy autonomous-fleet use could produce far greater disruption. The November 2025 storm occurred after the main U.S. planting season, so the paper’s authors describe a large agricultural loss as a counterfactual rather than an observed outcome.

New Space Economy coverage of the May 2024 tractor disruptions and its later solar-storm analysis make a broader point: GNSS value is concentrated downstream. Most economic damage can occur in farms, transport networks, telecom systems, finance, surveying, and machine automation even though the GPS spacecraft continue operating normally. That is one reason positioning, navigation, and timing resilience has become a commercial market rather than a specialized engineering topic.

How Space Weather Turns Navigation Into an Infrastructure Risk

Positioning, navigation, and timing, often shortened to PNT, supports far more than route guidance. The U.S. Department of Transportation defines PNT as the combined ability to determine location and orientation, move toward a desired position, and maintain precise time. Communications networks, financial systems, power systems, transportation, surveying, agriculture, and autonomous machines can all depend on one or more of those capabilities.

That dependence creates correlated exposure. A geomagnetic storm can disturb a large geographic region and affect many users at once. The November 2025 research is important because it documents strong mid-latitude amplitude scintillation across much of the continental United States, rather than a single local receiver anomaly. A regional failure mode can stress fleet management, dispatch, traffic control, logistics, emergency response, correction networks, and automated systems during the same period.

PNT resilience policy has been moving in this direction for years. The Department of Transportation’s national PNT architecture explicitly addresses GPS limitations, accidental interference, intentional jamming, spoofing, and complementary sources of navigation and timing. The department has also identified autonomous vehicles as an emerging use case that needs resilient PNT. NIST PNT guidance similarly treats GPS disruption as a risk-management problem because satellite timing and positioning support transportation and other essential services.

Space weather differs from jamming and spoofing in cause but overlaps with them in operational effect. Jamming overwhelms a legitimate transmission with interference. Spoofing presents counterfeit navigation information. Ionospheric scintillation changes the propagation of an authentic satellite transmission. A receiver may experience loss of lock, increased uncertainty, biased position, or unstable precision. From the vehicle’s point of view, all three conditions can reduce trust in satellite-derived location, so resilient system design benefits from common detection and fallback logic even when the underlying physics differs.

New Space Economy’s PNT analysis describes a market forming around this requirement. Low Earth orbit navigation broadcasts, terrestrial transmitters, inertial navigation, multi-constellation receivers, authenticated timing, correction services, and local positioning networks address different failure modes. No single substitute eliminates every source of error. Space weather strengthens the economic case for layered navigation because it demonstrates that even authentic, government-operated GNSS broadcasts can become unreliable at the point of use without an attack on the satellites or the receiver.

What Resilience Measures Can Reduce the Exposure?

The most effective response is layered localization. An automated vehicle should be able to compare GNSS against independent sources and recognize when the satellite solution no longer agrees with the physical world. Cameras, lidar, radar, inertial measurement, wheel odometry, detailed maps, road geometry, local ranging, and alternative radio-navigation services can all contribute. The engineering objective is graceful degradation, where loss of one source reduces capability in a controlled way instead of producing a sudden incorrect maneuver.

Integrity monitoring belongs at the center of that design. Accuracy describes how close a position is to the truth. Integrity describes whether the system can detect when the position should not be trusted. A navigation system suitable for automated transport needs limits on acceptable uncertainty, alarms when those limits are exceeded, and a defined response. Depending on speed, road type, traffic density, and available sensors, that response might include reducing speed, increasing following distance, avoiding lane changes, leaving a highway, stopping at a safe location, or suspending autonomous service. The Department of Transportation’s PNT work treats integrity and resilience as engineering requirements for future transportation systems.

Space-weather information can become another input to operational decision-making. NOAA’s Space Weather Prediction Center issues geomagnetic watches, warnings, and alerts, and its scintillation guidance explains that severe ionospheric conditions can prevent GPS receivers from calculating reliable positions. NOAA’s SOLAR-1 observatory, formerly SWFO-L1, reached the Sun-Earth L1 region in January 2026 and moved into its operational phase after commissioning in 2026. Its upstream solar-wind measurements improve the observations available to forecasters before disturbances reach Earth.

Forecasting alone will not eliminate the problem because the local ionospheric response can be complex. The November 2025 paper calls for coordinated observations and physics-based modeling to improve prediction of radio-frequency effects. Vehicle systems also need local evidence from their own receivers. Measures such as monitoring carrier-to-noise ratios, satellite geometry, residual errors, multi-frequency consistency, correction quality, and disagreement with inertial or map-based localization can help identify degradation that a regional warning cannot specify at street level.

Independent PNT services are also moving closer to practical deployment. Iridium announced commercial PNT chip availability in July 2026 for authenticated timing and location using its low Earth orbit network. Xona’s Pulsar-0 demonstration satellite completed more than 350 transmission passes across four continents during its initial year in orbit, and TrustPoint has been testing its Low Earth Orbit Navigation System. These services have different designs and maturity levels, so they should be evaluated on measured performance, integration cost, coverage, integrity, and failure independence rather than treated as interchangeable replacements for GPS.

Why GPS Resilience Is Becoming a Space Economy Market

GNSS has long been treated as public infrastructure because governments operate the core constellations and make civil broadcasts broadly available. Commercial value accumulates around receivers, correction services, mapping, timing, fleet software, surveying, agriculture, mobile devices, and automation. As dependence grows, customers begin paying for reliability above the baseline service, and that creates a market for commercial PNT. New Space Economy’s applications guide describes how positioning and timing already feed numerous commercial and industrial activities.

Autonomous transport is a strong demand source because it places a premium on continuity and integrity. A smartphone can tolerate a temporary map error with little consequence. A machine making steering or routing decisions cannot assume that every available position fix is trustworthy. Commercial services can sell stronger radio links, independent timing, correction data, authentication, additional constellations, interference detection, inertial integration, and service guarantees. New Space Economy’s commercial PNT review describes a transition from single-system dependence toward layered service design for autonomy, telecom, finance, defense, and industrial users.

The November 2025 storm expands the buyer’s threat model. Many commercial PNT discussions focus on jamming and spoofing because those threats have become frequent in conflict zones and transportation corridors. Space weather demonstrates a separate route to the same business problem: trusted satellite transmissions can be degraded by the medium through which they travel. A backup that shares the same propagation path and frequency assumptions may fail at the same time. Buyers need to ask whether a backup is technically independent enough to remain useful under the event they are trying to survive.

This creates commercial room for combinations rather than one universal successor to GPS. Low Earth orbit PNT can provide stronger received radio power and different geometry. Terrestrial systems avoid the long satellite path. Inertial systems continue operating without external radio transmissions for limited periods. Map-relative localization can anchor a vehicle to local features. Multi-constellation GNSS can improve satellite geometry and availability, though a large ionospheric disturbance can affect broadcasts from several constellations because they cross the same atmosphere. Current commercial development by Xona, Iridium, and TrustPoint illustrates how providers are pursuing different technical approaches rather than a single replacement architecture.

The space economy consequence reaches beyond autonomous cars. Insurance, logistics, aviation, maritime operations, agriculture, mining, telecom timing, data centers, defense, and emergency services all have reasons to buy resilience. A severe geomagnetic storm can impose correlated losses across several of these sectors. That turns space-weather-aware PNT from a niche technical capability into a broader infrastructure service market.

Summary

The November 2025 geomagnetic superstorm produced a measured navigation problem, not a hypothetical one. The Geophysical Research Letters study published on August 29, 2026 found strong mid-latitude ionospheric scintillation across much of the continental United States and reported horizontal GNSS positioning errors exceeding 10 meters in affected regions. The same physical process connected intense auroral activity, steep electron-density gradients, radio-wave distortion, and degraded positioning.

That evidence supports concern about autonomous transportation, but the safety interpretation needs precision. The research does not report a self-driving vehicle crash caused by the storm. Modern high-automation systems use multiple sensors and maps, so a bad GPS fix should be only one input in a larger localization system. The real engineering test is whether the vehicle detects that the navigation estimate has degraded and whether it can continue safely or enter a minimum-risk condition.

The May 2024 Gannon storm shows why the issue cannot be dismissed as a rare scientific curiosity. GPS errors reached tens of meters in parts of the United States, precision agriculture systems were disrupted during planting, and NOAA later cited more than $500 million in potential profit losses. Two severe storms in consecutive years produced documented GNSS degradation through related ionospheric mechanisms, even though their timing and regional effects differed.

The response is broader than better GPS reception. Autonomous systems need sensor diversity, integrity monitoring, safe fallback behavior, space-weather awareness, and positioning sources with enough technical independence to avoid common-mode failure. Government PNT policy already points toward complementary services, and commercial providers are bringing low Earth orbit and terrestrial alternatives into the market.

For GPS and self-driving cars, the lesson from November 2025 is that location cannot be treated as a single number delivered by a satellite receiver. It is an estimate with uncertainty, dependencies, and failure modes. The safer path for autonomous transportation is to make that uncertainty visible to the machine and to design the vehicle so that losing confidence in satellite navigation reduces capability predictably rather than producing a confident mistake.