Introduction
The ionosphere and troposphere are the two largest sources of GNSS error. Understanding how they affect signals, and how we model them, is key to achieving high accuracy.
The Two Layers
- Ionosphere: 60–1000 km altitude; charged particles (ions, electrons); dispersive (affects different frequencies differently); varies with day/night and solar cycle
- Troposphere: 0–18 km altitude; neutral atmosphere (gases, water vapour); non-dispersive (affects all frequencies equally); weather-dependent
Ionospheric Effects
Signals are delayed as they interact with free electrons. Delay amount: 1–15 metres (varies dramatically). Key factors: time of day (peak after local noon), solar activity (11-year cycle), latitude (worst near equator and auroral zones), and elevation angle (lower satellites = longer path = more delay).
Dispersive property: Ionospheric delay is proportional to 1/f², meaning signals on different frequencies experience different delays. This is the key to correcting it.
Correcting Ionosphere
- Dual-frequency (best): Compare L1 and L2/L5 arrival times to calculate exact delay, eliminates ~95% of ionospheric error
- SBAS: WAAS/EGNOS broadcast a grid of ionospheric delay values, corrects 50–70%
- Broadcast model: GPS message includes an 8-parameter model, corrects ~50%
- PPP: Estimates ionosphere as part of the solution (takes time to converge)
Tropospheric Effects
Signals are delayed by dry gases and water vapour. Delay amount: 2–25 metres at zenith (2–3x more at low elevations). Two components:
- Hydrostatic (dry): ~90% of delay, predictable from pressure
- Wet (water vapour): ~10% of delay, highly variable
Correcting Troposphere
The non-dispersive challenge: unlike ionosphere, you can't use dual-frequency to fix troposphere. Alternative methods:
- Models (Saastamoinen, Hopfield, GPT): Use pressure, temperature, humidity, accuracy 80–95% with good inputs
- Estimation: Treat as an unknown parameter in PPP, takes time to converge
- Differential (short baselines <10 km): Troposphere cancels naturally; longer baselines need modelling
Advanced Atmospheric Products
- NWP (Numerical Weather Prediction): Uses weather forecast data for very accurate troposphere corrections, used in PPP services
- VMF1 (Vienna Mapping Functions): Based on ECMWF data for high-accuracy troposphere models
- IONEX (Ionosphere Exchange): Global ionospheric maps produced from GNSS networks, available for post-processing
Practical Impact
| Technique | Ionosphere Error | Troposphere Error |
|---|---|---|
| Single-frequency, no model | 5–15 m | 2–5 m |
| Single-frequency + SBAS | 2–5 m | 1–3 m |
| Dual-frequency | 0.1–0.5 m | 1–3 m |
| Dual-frequency + model | 0.1–0.5 m | 0.2–0.5 m |
| RTK (<10 km baseline) | Cancels | Cancels |
| PPP (converged) | Estimated | Estimated |
Vital Points
- Ionosphere and troposphere are the largest error sources in GNSS
- Ionosphere is dispersive → fix with dual-frequency
- Troposphere is non-dispersive → fix with models or estimation
- Short-baseline RTK cancels both automatically
- PPP estimates both as part of its solution
- Understanding the atmosphere is key to achieving high accuracy