Why GPS Is Not Always Accurate

The GPS accuracy figures published by GPS.gov — 4.9 m open sky, 1–2 m with SBAS — describe conditions that don't always hold. In a real-world day, a smartphone GPS reading can be off by metres or even hundreds of metres for reasons that have nothing to do with the satellite constellation working as specified. This article covers what goes wrong, why, and what you can do about it.

The /learn/gps-accuracy-explained companion covers the best civilian GPS achieves. This one covers the failures.

The dominant failure mode: multipath

In open environments, GPS signals reach the receiver via a direct line-of-sight path from each satellite. In urban environments, those signals also reach the receiver via reflections off buildings, glass facades, water, ice, and other surfaces. The reflected signals arrive a few nanoseconds to a few microseconds later than the direct signal, depending on the reflection geometry.

The receiver, trying to compute a pseudorange from the satellite signal's arrival time, sees a delayed signal — which makes the satellite appear farther away than it really is. The result: the computed position shifts toward whatever direction the reflections are coming from. In dense urban environments with tall buildings on one side and open sky on the other, this can introduce systematic position errors of 50+ metres.

Quantitatively, the multipath contribution depends on:

• Surface reflectivity at the GPS frequency (~1.5 GHz). Glass and metal-clad buildings are particularly bad; concrete and brick are less so.
• Reflector distance and angle. A building 50 m away produces ~50 ns of delay; one 500 m away produces ~500 ns.
• Receiver design. Higher-end receivers use multipath- rejection algorithms (correlator design, carrier-phase smoothing) to identify and de-weight reflected components.

Modern smartphones partly mitigate multipath through software techniques and (in dual-frequency devices) by comparing L1 vs L5 timing, but no consumer receiver eliminates it. Dense urban environments are simply hard.

Indoor positioning: no signal

GPS signals are extraordinarily weak when they reach Earth. The broadcast power is ~25 W from a satellite at 20,200 km altitude; the received power at ground level is around −155 dBW (about 3 × 10⁻¹⁶ watts). That's already weaker than the thermal noise in a typical receiver; modern GPS works by sampling and correlating the signal over many milliseconds to extract it from below the noise floor.

Building walls attenuate the L-band signal substantially:

• Wood frame: 5–15 dB attenuation. Marginal indoor GPS may work near windows.
• Brick / concrete: 15–30 dB attenuation. GPS effectively doesn't work indoors.
• Metal-framed (commercial buildings): 30+ dB. GPS unusable.
• Underground / submarine: complete signal loss.

Indoor positioning therefore uses different technologies:

• Wi-Fi triangulation. The phone's OS measures signal strength from known Wi-Fi access points and triangulates. Accuracy: 5–50 m in well-mapped urban areas.
• Bluetooth beacons (Apple iBeacon, Google Eddystone). Pre-installed transmitters in shopping centres, airports. Accuracy: 1–5 m.
• Ultra-wideband (UWB). Premium smartphones (iPhone 11+) use UWB for centimetre-level indoor positioning near cooperative transmitters.
• Inertial dead reckoning. The phone's accelerometer and gyroscope estimate motion from the last known GPS fix. Drifts at ~10 m/min without correction.

The Coordinately /tools/my-location uses the browser's navigator.geolocation API, which combines all of the above. The returned accuracy estimate reflects whichever method was used.

Solar storms and ionospheric disturbances

Earth's ionosphere — the upper atmosphere's electrically charged layer — bends and slows GPS signals. Under normal conditions, receivers correct for this using broadcast ionospheric models (single-frequency) or by direct measurement via dual-frequency observations.

During solar storms, the ionosphere becomes much more variable and the corrections fail. A major event like:

• October 2003 Halloween Storms — degraded GPS accuracy worldwide; FAA WAAS service interrupted in some sectors.
• August 2017 (Hurricane Irma timing) — solar flare coincided with the hurricane; degraded emergency-response GPS for hours.
• May 2024 G5 storm — strongest in 20 years; degraded GPS in CONUS for ~48 hours; aviation users reported widespread WAAS unavailability.

The NOAA Space Weather Prediction Center issues real-time forecasts and warnings. High-precision users monitor space weather; consumer users notice mostly via app- reported “poor GPS” warnings.

Atmospheric and weather effects

Lesser but real contributions:

• Tropospheric delay — the lower atmosphere's water vapour and temperature variations affect signal propagation. Receivers model the effect; sub-metre residual error typical.
• Heavy rain and snow — minor signal attenuation; usually not noticed in positioning, but can degrade SBAS reception on certain weather radar bands that overlap with the GPS L-band.
• Solar elevation effects — at high latitudes during summer, the satellite geometry can be poorer because all visible satellites are concentrated to the south.

Jamming and spoofing

In recent years, deliberate interference with GPS has become a much more visible threat:

Jamming is the transmission of high-power noise on the GPS frequency, drowning out the legitimate signal. It's criminalised in most countries (in the US, FCC enforcement actions against jammer-using truck drivers happen multiple times per year), but jammers are cheap (~$50 online) and trivially used. Reported jamming sources include:

• Truck drivers blocking employer GPS trackers (common in the US).
• Military operations and military exercises.
• Geopolitical disruption (e.g., reports from war zones since 2022 of widespread regional jamming).
• “Personal privacy” jammers, illegal in most jurisdictions.

Spoofing is harder — it requires transmitting fake satellite signals that look legitimate. Documented spoofing incidents:

• Maritime spoofing in the Black Sea (2017–) — ships repeatedly report their GPS positions inland, near airports.
• Aviation spoofing — increasing reports since 2023 of commercial aircraft GPS being spoofed in the Middle East, forcing pilots to navigate by inertial systems.
• Sub-metre research demonstrations using software-defined radios.

Civilian receivers have no built-in spoofing protection. Modernised military receivers (M-code) do, via cryptographic authentication; civilian L1C and L5 include features that make spoofing harder but not impossible. Commercial users increasingly install RAIM (Receiver Autonomous Integrity Monitoring) and cross-check GPS against inertial or other sensor systems.

Antenna and receiver issues

The receiver itself can be the bottleneck:

• Antenna obstruction. Phone in a pocket, watch under a sleeve, GPS unit inside a backpack — all attenuate the signal. Performance improves dramatically with the antenna in clear view of the sky.
• Antenna orientation. Smartphones have a horizontal- patch antenna that's optimised for the phone lying flat or held upright; a phone lying screen-down on a dash may have reduced sensitivity.
• Old hardware. A 2010-era GPS chip is much less sensitive than a 2024-era one. Modernised receivers handle weaker signals (urban canyons, near-windows indoor) much better than older designs.

Mitigations

For consumer users who care about better GPS readings:

1. Get the latest hardware. Dual-frequency L1+L5 makes a real difference. Premium phones since ~2020 are substantially better than 2015-era hardware.
2. Step outside. Open sky is the single largest accuracy improvement. Indoor + multipath is the worst case.
3. Wait 30+ seconds after starting. The receiver accumulates measurements over time; the first fix is usually less accurate than the 30-second-old fix.
4. Enable Location Services with high accuracy. Some devices have a “battery saver” GPS mode that uses cellular triangulation only; switch to high accuracy for actual GPS.
5. Move the phone outside metal objects. A car's metal roof attenuates GPS; a phone on the dashboard near a metal pillar is worse than a phone in an open seat.
6. Use SBAS-aware devices. Most modern phones include it by default; some pre-2015 devices don't.

For survey-grade users, the answer is RTK or PPP with a proper dual-frequency receiver and base station — not a smartphone.

Real-world degradation examples

A few specific environments and their typical GPS-accuracy budgets:

• Manhattan, NYC midtown: 20–100 m horizontal error from multipath off skyscrapers; readings frequently jump 50 m as the multipath geometry shifts with movement.
• Tokyo's Shibuya district: similar urban-canyon effects; 30+ m errors are common.
• Inside a car on a highway: 5–15 m; the metal roof attenuates the signal somewhat, but the phone's antenna near the windscreen usually still works.
• National park trails with dense canopy: 10–30 m; attenuation by tree cover plus multipath off trunks. Hiking GPS apps deliberately use elevation-aware path-matching to smooth this.
• Indoor shopping mall: GPS effectively unavailable; positioning falls back to Wi-Fi triangulation, typically 10–50 m accuracy.
• Cruise ship at sea: open-sky conditions, 5 m typical; but the steel ship can attenuate signal in interior cabins.
• Polar research station, summer: 5 m typical; the satellite geometry is good but the ionospheric variability is higher than mid-latitudes.

The pattern: the GPS itself works fine in all of these environments. The accuracy budget depends overwhelmingly on the receiver's view of the sky and the multipath environment around it.

Common misconceptions

“GPS is jamming-resistant.” Civilian GPS is not. Jamming has been demonstrated by amateur radio operators using software-defined radio; military jammers are far more capable. Anti-jamming requires antenna arrays (commercial CRPA systems) or cryptographic authentication (military M-code), neither of which is in consumer devices.

“GPS errors are random and average out.” Some errors (receiver thermal noise) are random; many (multipath, ionospheric, ephemeris) are systematic and don't average out. Sitting still and waiting for a fix to “converge” helps with the random component but not the systematic.

“GPS uses encryption.” Civilian GPS does not. Military GPS does (the P(Y) code is encrypted). The lack of authentication in civilian GPS is the structural reason spoofing is technically possible.

“Cellular triangulation is more accurate than GPS in cities.” Cellular triangulation gives 50–500 m accuracy. GPS in urban canyons gives 10–50 m. So cellular is slightly better only in the worst urban canyons; in most urban environments, GPS is still more accurate than cellular. Modern devices fuse all available signals to produce the best estimate.

“If my GPS says I'm in the wrong place, the GPS is broken.” Almost always, the environment is the problem — multipath, attenuated signal, ionospheric disturbance, or jamming. The satellites themselves are working as specified ~99.99% of the time.

“GPS works the same everywhere on Earth.” Polar regions have somewhat different satellite geometry (no satellites pass directly overhead because of the 55° inclination). Near the magnetic equator, ionospheric variability is higher. Mid-latitudes have the best baseline conditions; high-latitude and equatorial regions degrade slightly.