Why Arctic GNSS Positioning Fails Above 70°N – And What Actually Works

The Problem No One Wants to Admit

Positioning hardware proven in the North Sea does not automatically perform the same way above 70°N. The physics changes, and so does the failure mode. At high latitude, dynamically positioned vessels can lose positioning integrity during geomagnetic disturbances, survey corrections drop out for extended periods, and inertial systems coast for far less time than their mid-latitude track record suggests. These are not edge-case theories; they are recurring operational realities, and the available standards say very little about them.

Three mechanisms drive the degradation, and all three are well established in the geodesy and space-weather literature. First, geometry. GPS satellites sit in orbital planes inclined at roughly 55° to the equator, so they never pass directly overhead at high latitude. Above the Arctic Circle (≈66.5°N) the highest satellite stays well down toward the horizon, and at the pole the best available elevation is only around 45°. The result is better horizontal but markedly worse vertical geometry, which inflates vertical dilution of precision and degrades height accuracy. PPP and RTK convergence is a separate problem driven mainly by ionospheric activity and ambiguity resolution, as the next section explains. Second, the ionosphere. The auroral oval and polar cusp sit directly over these operating areas, and the scintillation they produce attacks GNSS carrier phase. Third, the redundancy you rely on at mid-latitude – SBAS, dense RTK base networks, reliable real-time corrections – simply thins out or disappears the further north you go.

What Actually Happens at High Latitudes

Ionospheric scintillation is the dominant high-latitude failure mode, and it does not affect both frequencies equally. Cycle slips occur overwhelmingly on the L2 band, because scintillation is stronger at lower frequencies and L2 is more vulnerable than L1. During disturbed conditions, phase scintillation and cycle slips concentrate on the dayside cusp and in the nightside auroral oval – exactly the regions that sit over the High Arctic. When scintillation is severe enough to corrupt both L1 and L2, dual-frequency receivers lose the very combination they depend on to fix ambiguities, RTK ambiguity-resolution success rates fall and time-to-fix lengthens.

This is documented field behaviour, not modelling. Research on precise GNSS positioning in Greenland has shown that ionospheric disturbances measurably degrade RTK and PPP performance at these latitudes. The Norwegian Mapping Authority (Kartverket) operates the nationwide network-RTK service CPOS, and studies around Tromsø (≈70°N) have found CPOS positioning severely disturbed for hours even during relatively modest geomagnetic events. The clearest recent example is the St. Patrick’s Day storm of 17–18 March 2015 – the most intense geomagnetic storm of Solar Cycle 24, reaching Kp = 8 (G4, severe) – which produced strong GNSS disturbances across Norway, with the northern parts of the country worst affected and disturbed longest. Severe storms do not just nudge accuracy; they can render precise GNSS unusable for hours at a time, and at offshore day rates that is expensive downtime.

The INS Can’t Save You

When GNSS drops out, the standard fallback is inertial coasting. That works well in the North Sea, where outages are typically seconds long. The Arctic problem is duration: scintillation events can degrade positioning for minutes, and severe storms for hours, which is far beyond what most inertial systems can bridge while holding usable accuracy.

How far an inertial navigation system coasts depends almost entirely on its grade. Navigation-grade fibre-optic-gyro (FOG) and ring-laser-gyro (RLG) systems hold position to roughly a metre for a minute or two and degrade gracefully from there; tactical-grade MEMS units accumulate position error far faster and are typically only good for tens of seconds before they breach DP-relevant tolerances. The practical consequence is blunt: for dynamic positioning that needs 1–2 m accuracy, even a navigation-grade INS buys you a few minutes of coast, and a tactical MEMS unit buys you well under a minute. A ten-minute scintillation outage will exceed that envelope unless an independent acoustic reference is already online.

Heading is a second, separate problem. Conventional north-seeking gyrocompasses derive heading from the horizontal component of Earth’s rotation, and that component shrinks toward the poles – so gyrocompass heading error grows with latitude and becomes unreliable in the High Arctic. FOG and RLG systems are far less affected and should be treated as the reference, not the conventional gyrocompass. Temperature is a third factor that is easy to underestimate: moving sensors between a heated interior and an exposed deck imposes large thermal swings that shift MEMS bias, so any drift budget assumed in a warm cabin needs to be re-validated in operating conditions.

Neural Networks Are Not the Answer (Yet)

Machine-learning aiding of INS during GNSS outages is an active research area. Hybrid architectures that pair a Temporal Convolutional Network with an LSTM to generate pseudo-GNSS measurements, then fuse them through an adaptive Kalman filter, have reported meaningful error reductions over conventional methods during simulated outages of tens to hundreds of seconds.

The reported numbers are encouraging, but the limitations are decisive for operations. Results typically come from a single dataset and a single sensor; change the IMU and the network generally has to be retrained. Real-time performance on embedded hardware is rarely demonstrated, degradation of the heading reference is often ignored, and there is little independent third-party validation against navigation-grade reference truth. In short, this is a promising research direction, not an operational solution. Treat it as something to watch over the next several years, not as a substitute for proven inertial and acoustic backup today.

What the Standards Do Not Tell You

The authoritative GNSS positioning reference for this industry is the joint IOGP/IMCA publication Guidelines for GNSS Positioning in the Oil and Gas Industry (IOGP Report 373-19, also issued as IMCA S 015; second revision published 2021). It covers positioning principles, error sources such as multipath and atmospheric effects, and the statistical testing and quality measures needed for rigorous quality control. What it does not do is treat the polar case as special: ionospheric effects are addressed generically, as if a single mitigation strategy applies everywhere. It is worth being precise about what each document is – IMCA S 022, frequently cited in this context, is An Introduction to Inertial Navigation Systems, not the GNSS positioning guide, and neither document sets out polar-specific GNSS requirements.

That gap matters because the standards implicitly assume mid-latitude conditions: dense RTK base-station coverage and working SBAS augmentation. Neither assumption holds in the far North. None of the operational SBAS systems – EGNOS, WAAS, the others – provides meaningful service at Arctic latitudes, both because there are too few reference stations and because the geostationary augmentation satellites sit very low on the horizon (only a few degrees of elevation by 75°N). RTK base networks thin out and then vanish, real-time corrections over marginal cellular and satellite links become unreliable, and PPP convergence is exactly what high-latitude scintillation degrades. In practice, operators have to engineer their own Arctic positioning assurance.

What You Actually Need

The following is the configuration we would specify for DP, survey and subsea work north of 70°N. It is deliberately conservative, because the failure modes above are real and the standards do not yet cover them.

1. Deploy Navigation-Grade FOG or RLG INS

Tactical MEMS is not enough for safety-critical Arctic DP or survey. Specify a navigation-grade inertial system – the class represented by ring-laser-gyro units such as the Honeywell HG9900 or fibre-optic-gyro systems such as the iXblue/Exail PHINS – rather than a low-cost tactical MEMS module. Whatever the grade, validate coast performance before mobilising: disable GNSS and measure actual drift at 60, 120 and 300 seconds in operating conditions. Use the measured drift, not the datasheet, to set your coast tolerances. A well-behaved navigation-grade unit should hold close to a metre over the first minute, as noted above; the point of the test is to catch units that do not. To take a deliberately pessimistic illustration, if you need 2 m accuracy and a degraded or poorly aligned unit drifts 5 m in 60 seconds, your real coast time is well under a minute, not the few minutes the grade nominally implies.

2. Multi-Constellation, Multi-Frequency GNSS

Use all available constellations – GPS, GLONASS, Galileo and BeiDou – to recover satellite availability and geometry that GPS alone cannot provide at high latitude. GLONASS and Galileo carry higher orbital inclinations than GPS, which helps marginally with high-latitude geometry. Multi-frequency tracking (L1/L2/L5) is essential for ionospheric handling, and where receivers offer firmware tuned for high-latitude scintillation, use it. During severe storms, expect satellites to drop regardless, and plan an alternative.

3. Pre-Deploy an LBL Acoustic Array

For safety-critical work north of 70°N, do not rely on USBL alone as the GNSS fallback. Pre-install a long-baseline (LBL) acoustic array – several seabed transponders, calibrated against GNSS while it is available – so that a tested independent reference is already online when GNSS degrades. Configure automatic fallback to LBL during outages and recovery to GNSS afterwards, and leave the array in place for the duration of the campaign. Mobilising, calibrating and recovering an LBL array is a real cost, but it is far smaller than a lost DP day.

4. Integrate Space-Weather Monitoring

Subscribe to operational alerts from the NOAA Space Weather Prediction Center and equivalent European services, and monitor the Kp index and rate-of-TEC-index (ROTI) in real time. Fold the forecast into the daily positioning risk assessment: escalate contingency planning as Kp rises, and do not start critical DP operations when a severe geomagnetic storm is forecast in the next 12–24 hours. High-latitude GNSS degradation correlates strongly with geomagnetic activity, and the indices are freely available – there is no excuse for being surprised.

5. Configure Automated Failover

DP and navigation systems should switch from GNSS to acoustic/INS automatically when defined thresholds are breached. As illustrative starting values to tune against your own data: GDOP above 6, fewer than five tracked satellites, a cycle-slip rate sustained above roughly one slip per satellite per minute, or correction latency beyond 5 seconds. Test the switchover monthly by simulating a GNSS dropout and confirming the system transitions without operator intervention. Manual switching is exactly what fails at 3 a.m. in January with a storm warning live.

6. Dual Heading Reference

A single GNSS heading source is not enough at high latitude. Pair a dual-antenna GNSS heading source with a FOG-based heading reference and compare them continuously, because conventional gyrocompasses degrade with latitude and GNSS heading is exposed to multipath and low-elevation error. If no FOG heading reference is fitted, fit one. Heading loss during DP operations is far more costly than the install.

The Metrics That Matter

Track these, document them, and test them routinely:

• Position availability: target 99.5% for DP, 99.0% for survey, measured per operational shift (or rolling 24 hours) against an independent reference such as the LBL array, with availability defined as the fraction of epochs meeting the stated accuracy budget. Log every outage longer than 10 seconds and analyse the cause. If you cannot meet these, you are not ready for the Arctic.
• Average and 95th-percentile GDOP: persistently high values indicate weak satellite geometry for your area and operating window.
• Cycle-slip rate: a sustained rise signals worsening ionospheric conditions – monitor continuously, not just after the fact.
• INS coast performance: test monthly by disabling GNSS for 60, 120 and 300 seconds and comparing drift against specification. A large deviation from spec means an INS problem.
• Acoustic system latency: monitor update rate and position age; rising latency points to transponder or sound-velocity-profile issues.
• Kp / position-availability correlation: log them together. Over 6–12 months you build a correlation record specific to your operations, and you can tune your risk assessment to it.

What Standards Should Require

IOGP and IMCA have not published positioning guidance specific to polar regions. Reasonable additions would be:

1. Space-weather monitoring – Kp index, ROTI, solar-wind speed – recommended north of the Arctic Circle, with contingency planning triggered at severe-storm thresholds.
2. Minimum INS class for DP and survey north of 70°N – navigation-grade FOG for DP2/DP3 vessels, high-grade tactical FOG as a floor for survey vessels, with coast performance defined rather than assumed.
3. Independent acoustic positioning for safety-critical operations north of 70°N – LBL preferred, USBL only with proven redundancy.
4. GNSS/INS coast-performance testing before Arctic entry, with drift tolerances defined at 60, 120 and 300 seconds by operation type.
5. Multi-constellation, multi-frequency GNSS above 70°N – GPS plus GLONASS plus Galileo at minimum, dual-frequency standard and triple-frequency recommended.

None of this exists in current guidance. Until it does, operators have to write their own Arctic positioning procedures – and the safe assumption is that mid-latitude performance does not transfer north.