GNSS Errors Mapped: Why a Perfect Satellite Still Lands You Off by Meters
TL;DR
- A GNSS fix drifts because errors enter at several points along the signal path, from the satellite clock to the wall beside your antenna.
- Six error sources explain most of the meter-level budget: satellite clock error, ephemeris error, ionospheric delay, tropospheric delay, multipath, and receiver noise.
- The numbers in a practical error budget are residuals after basic broadcast corrections, not the raw delays before the receiver applies its standard models.
- Dual-frequency GNSS removes most first-order ionospheric delay because the ionosphere affects different frequencies by different amounts.
- Errors far from the antenna are widely shared; local reflections are not. That distinction explains both the value and the limits of later differential methods.
GNSS error sources can be mapped along one journey: the signal leaves a satellite about 20,000 km above Earth, crosses two layers of atmosphere, and reaches an antenna surrounded by its own local environment.
Open a map on your phone and stand still. The blue dot will usually move a little anyway. It may drift across the pavement, jump toward the other side of the road, then settle back.
That static drift is the accumulated result of ranging errors introduced while GNSS signals travel roughly 20,000 km from space to the receiver. The signal crosses several physical layers before it reaches the antenna, and each layer adds a different kind of measurement error.
In How GNSS Works, we reduced positioning to one idea: distance equals the speed of light multiplied by signal flight time. Any factor that makes that flight time look wrong becomes a ranging error. This article follows the signal down to the antenna, marks where each error enters, estimates how large it is, and then looks at which parts can be reduced.
Need-to-Know Terms
Five terms organize the rest of the error map:
- Ephemeris — The orbital data used by the receiver to calculate where a satellite was when it transmitted a signal.
- Ionosphere — A region of charged particles roughly 50–1,000 km above Earth that changes signal propagation time.
- Troposphere — The lower atmosphere where pressure, temperature, and water vapor add another delay.
- Multipath — A measurement error caused when reflected copies of a signal reach the antenna after traveling longer paths.
- UERE — User Equivalent Range Error, a way to combine several ranging-error contributions into one practical budget.
The Signal's Journey: Three Zones, Three Kinds of Error
GNSS errors enter in three physical zones: the satellite, the atmosphere, and the local environment around the antenna.
The three zones can be summarized as follows:
| Zone | Where It Sits | What Enters the Measurement |
|---|---|---|
| The satellite | The signal source in space | Residual timing error and residual orbital-position error. |
| The atmosphere | The path between space and the receiver | Propagation delay through charged particles and neutral air. |
| Your surroundings | The final meters around the antenna and receiver | Reflections, blocked line-of-sight paths, and receiver measurement noise. |
The closer an error source is to the antenna, the less likely it is to be shared by another receiver.
Engineer's Takeaway
GNSS errors follow a spatial pattern: satellite-side errors are widely shared, atmospheric delays remain regionally correlated, and multipath is local to one antenna environment. That pattern determines which errors later correction methods can remove.
Space-Segment Errors: Satellite Clock and Ephemeris
The first two errors are already present before the signal reaches the atmosphere. Both originate at the satellite and both affect a broad area in nearly the same way.
Satellite Clock Error
GNSS depends on time. Satellite atomic clocks are extremely stable, but they still drift by small amounts. A timing error becomes a distance error immediately: as How GNSS Works showed, one microsecond multiplied by the speed of light is about 300 meters.
The receiver does not work with an uncorrected satellite clock. GNSS control stations track each satellite and the navigation message includes clock-correction parameters. After those basic corrections are applied, a practical residual contribution is about ±2.0 m.
Ephemeris Error
A satellite also tells the receiver where it is. More precisely, it broadcasts orbital parameters that let the receiver calculate its position at transmission time. Those parameters are predictions maintained by the ground-control system. The predicted orbit and the real orbit never match perfectly.
If the satellite position is slightly wrong, the calculated range to that satellite is slightly wrong as well. After the standard broadcast update process, the residual contribution is typically about ±2.5 m. Like satellite-clock error, it remains broadly shared across nearby receivers, although the correlation weakens as the receivers move farther apart.
Atmospheric Errors: Ionospheric and Tropospheric Delay
The signal next crosses two physically different parts of the atmosphere. The ionosphere is a dispersive layer of charged particles. The troposphere is neutral air shaped by pressure, temperature, and water vapor. Both alter the measured flight time, but not in the same way.
Ionospheric Delay
The ionosphere extends roughly 50–1,000 km above Earth. Solar radiation creates free electrons in this region. When GNSS signals pass through it, their propagation time changes and the receiver interprets the delay as extra range.
The delay is not constant. It varies with location, time of day, season, and solar activity. Daytime conditions are usually stronger than night-time conditions, and periods of elevated solar activity can raise the delay substantially. A typical residual contribution for an error budget is about ±5.0 m; during strong ionospheric activity, the effect can exceed 15 m.
The important physical property is frequency dependence. The ionosphere changes different GNSS frequencies by different amounts. That is the reason dual-frequency receivers can estimate and remove most first-order ionospheric delay later in the processing chain.
Tropospheric Delay
The signal then crosses the lower atmosphere. Pressure, temperature, dry gases, and water vapor all change the propagation path slightly. Unlike the ionosphere, the troposphere is effectively non-dispersive at GNSS frequencies: two GNSS frequencies see almost the same delay.
Most of the effect comes from the hydrostatic, or dry, component. It is relatively predictable. The smaller wet component changes with local weather and is harder to model. After standard modeling, a practical residual contribution is about ±0.5 m. Low-elevation satellites are harder because their signals travel obliquely through a thicker section of atmosphere, so the remaining error can become several meters.
Local Environment Errors: Multipath and Receiver Noise
The final two error sources enter close to the antenna. They are local rather than regional. A second receiver only a short distance away may see a different wall, a different ground reflection, and a different noise floor.
Multipath Error
A GNSS signal does not always arrive by one direct path. It can reflect from the ground, a building facade, a vehicle, or nearby equipment before reaching the antenna. The receiver then sees a direct signal and one or more delayed copies. Those copies interfere with the ranging measurement.
A more severe case occurs when the direct line of sight is blocked and the receiver relies on a reflected non-line-of-sight path. The reflected path is longer, and that extra distance enters the pseudorange. In open-sky conditions, a practical error-budget contribution is about ±1.0 m. In dense urban canyons, multipath and NLOS reception can push position errors into tens of meters.
Receiver Noise
The receiver itself adds a smaller floor: thermal noise, quantization, tracking-loop limits, and other electronic effects introduce measurement variation. Modern high-quality code measurements can reach about ±0.3 m of receiver-noise contribution, although consumer-grade code measurements may be higher.
Receiver noise matters because it cannot be shared or corrected from far away. It is usually smaller than atmospheric error or urban multipath, but it defines the floor that remains after the larger effects are reduced.
Mapping the Full GNSS Error Budget
Once the six sources are placed in one table, the earlier spatial pattern becomes visible. The values below are residual contributions after basic broadcast corrections and standard modeling. They are not raw, uncorrected propagation delays:
| Error Source | Zone | Scope | Typical Residual |
|---|---|---|---|
| Satellite clock error | Space | Global — nearly common across a region | ±2.0 m |
| Ephemeris error | Space | Global — broadly shared, then decorrelates with distance | ±2.5 m |
| Ionospheric delay | Atmosphere | Regional — similar over a limited area | ±5.0 m |
| Tropospheric delay | Atmosphere | Regional — similar over a limited area | ±0.5 m |
| Multipath | Local | Local — specific to the antenna environment | ±1.0 m in open sky; much higher in urban canyons |
| Receiver noise | Local | Local — specific to the receiver measurement chain | As low as ±0.3 m for modern receivers |
The largest normal contribution in this table is ionospheric delay. The two satellite-side terms are also substantial, but they have one useful property: receivers in the same area see nearly the same error. Local multipath and receiver noise are smaller in open sky, yet they are harder to share or predict.
These six values should not be added directly. A simple sum would assume that every error reaches its worst value in the same direction at the same moment. That is not how independent error terms behave. Engineering budgets normally combine them by squaring each contribution, adding the squares, and taking the square root. Using the open-sky multipath value above gives an equivalent ranging error of roughly 6 m.
That is still not the final map-position error. The conversion from ranging error to position error depends on how the satellites are distributed across the sky. Widely spread satellites constrain the solution better than satellites clustered in one direction. That geometry factor is DOP. GNSS Accuracy Decoded covers DOP and accuracy metrics in detail.
Engineer's Takeaway
The budget shows two things at once: ionospheric delay is usually the largest normal residual, and error sources sit on a spectrum from shared to local. A practical correction method works well only when it matches the physical layer where the error originates.
Which Errors Can Be Removed?
The mitigation strategy depends on the physical layer where each error originates.
The six error sources require different correction methods and leave different residual limitations:
| Error Source | Typical Residual | Primary Mitigation | Remaining Limitation |
|---|---|---|---|
| Satellite clock error | ±2.0 m | Broadcast clock parameters; augmentation corrections | Residual clock bias remains |
| Ephemeris error | ±2.5 m | Broadcast ephemeris; precise orbit corrections | Broadcast orbit is still a prediction |
| Ionospheric delay | ±5.0 m | Dual-frequency ionosphere-free combination | Higher-order and disturbed-ionosphere residuals remain |
| Tropospheric delay | ±0.5 m | Atmospheric model; elevation weighting or mask | Wet delay remains difficult to predict |
| Multipath | ±1.0 m in open sky | Antenna placement; multipath-resistant antenna design | Local reflections cannot be fully modeled |
| Receiver noise | As low as ±0.3 m | RF design; filtering and smoothing | Defines the receiver-side noise floor |
Satellite Clock and Ephemeris Errors
Receivers already remove most satellite clock and orbit error as part of normal operation. The navigation message carries clock-correction parameters and broadcast ephemeris. The receiver decodes them and applies the corrections before calculating position. The ±2.0 m and ±2.5 m values above are the remaining residuals, not the original raw errors.
Augmentation systems such as SBAS can transmit additional clock, orbit, and ionospheric corrections over a wider region. That pushes the remaining satellite-side contribution lower without changing the basic receiver architecture.
Ionospheric Delay
Ionospheric delay is where dual-frequency GNSS pays off directly. Because the delay depends on frequency, a receiver can compare measurements from two frequencies and calculate most of the first-order ionospheric term. ESA Navipedia notes that a dual-frequency combination can remove more than 99.9% of the first-order effect.
A single-frequency receiver must use a broadcast model instead. The GPS Klobuchar model is useful, but it is still an approximation: it is generally estimated to remove about half of the RMS ionospheric range error. The signal strategy covered in GNSS Signals Explained matters because the value of a second frequency is realized here.
The distinction is physical: the ionosphere is dispersive, so a dual-frequency receiver can solve for most of the first-order delay algebraically.
Tropospheric Delay
Tropospheric correction does not use the same frequency comparison. Neutral-air delay is effectively non-dispersive at GNSS frequencies, so its residual must be modeled or estimated. The dry component is relatively predictable; the wet component changes with local weather and remains harder to estimate. Low-elevation satellites also travel through a longer atmospheric path, so receivers often apply elevation weighting or a cutoff mask.
Pro Tip
A cutoff angle around 5°–10° is a practical starting point when low-elevation measurements are bringing excessive atmospheric path length and multipath into the solution. The correct threshold depends on satellite availability and the installation environment.
Multipath
Multipath is mainly an installation problem. Better receiver processing and antenna design can reject part of it, but no general model can predict the exact reflection created by a nearby wall, roof edge, vehicle, or wet ground surface. The first actions are physical: move the antenna away from reflective structures, preserve a clear sky view, and use an antenna designed for the operating environment.
Receiver Noise
Receiver noise is reduced with hardware design and signal processing: a controlled RF chain, low-noise components, stable tracking loops, digital filtering, and smoothing across measurements. It is not usually the largest error source, but it becomes more visible after the larger terms are reduced.
At this point the boundary is clear. A receiver can correct most satellite-side error, remove most first-order ionospheric delay with two frequencies, model part of the troposphere, and control its own measurement noise. Multipath remains strongly local. To move from meter-level positioning toward centimeters, the next step is to introduce a second receiver and remove shared residuals through differential processing. A later RTK guide will explain that method in detail.
Conclusion
Static position drift is the combined result of six error sources across the space segment, the atmospheric path, and the local receiver environment. Two originate at the satellite, two enter during propagation, and two arise close to the antenna.
That map explains three practical facts. GNSS stays at the meter level because the residual budget is still several meters. Dual-frequency reception matters because ionospheric delay is often the largest normal contribution and it is frequency-dependent. Further improvement needs another reference point because the remaining errors do not all yield to modeling inside one receiver.
A later RTK guide will start from that second receiver and explain how the same measurement chain can be pushed toward centimeter-level positioning.
Key Takeaway
A GNSS fix lands at the meter level not because of one large error, but because six independent errors enter along the signal path from space to the device. A receiver can remove most satellite-side error and most first-order ionospheric delay on its own. Multipath remains strongly local, which is why the next step is a second receiver and why RTK exists.
Frequently Asked Questions
What causes GNSS positioning errors?
GNSS positioning errors enter at three stages of the signal path. Satellite clock error and ephemeris error originate in space. Ionospheric delay and tropospheric delay appear while the signal crosses the atmosphere. Multipath and receiver noise enter close to the antenna and receiver. A practical GNSS error budget combines the residual contribution from all six sources after the receiver has applied its normal broadcast corrections and models.
Why is my receiver less accurate during the day?
Daytime solar radiation increases ionization in the upper atmosphere, so ionospheric delay is often stronger during the day than at night. The effect also changes with latitude, season, and solar activity. A single-frequency receiver relies on a model and therefore leaves a larger residual. A dual-frequency receiver can compare two frequencies and remove most first-order ionospheric delay directly.
Why does dual-frequency GNSS remove ionospheric delay but not tropospheric delay?
The ionosphere is dispersive: it changes the propagation time of different GNSS frequencies by different amounts. A receiver can use that difference to estimate and remove most first-order ionospheric delay. The troposphere is effectively non-dispersive at GNSS frequencies. Both frequencies see nearly the same tropospheric delay, so frequency comparison cannot isolate it. Tropospheric delay must be modeled or estimated instead.
How do the six GNSS errors add up to a total?
The six GNSS error sources should not be added as a simple worst-case sum. They are partly independent and can be positive or negative. Engineering budgets normally combine their residual contributions with a root-sum-square method: square each value, add the squares, and take the square root. Using typical open-sky values gives roughly 6 m of equivalent ranging error. Final position error also depends on satellite geometry, described by DOP.
Can a GNSS receiver reach centimeter accuracy by itself?
A receiver can remove most satellite-side errors, estimate most first-order ionospheric delay when two frequencies are available, model part of the troposphere, and reduce its own noise. It still cannot fully remove local multipath or all residual shared errors by itself. Centimeter-level positioning normally adds a second reference receiver and differential processing. That is the starting point for RTK, which will be covered in a later guide.
You Might Also Like
GNSS Accuracy Decoded
Separate CEP, RMS, R95, and DOP before comparing accuracy claims across receivers.
LateralGNSS Signals Explained
See how constellations and L1/L5 signal choices shape the measurements available to a receiver.
ApplyKalmix SCOUT PRO
Evaluate dual-frequency RTK positioning with a field-ready USB-C GNSS receiver.
