Intelligent global navigation satellite system solutions for slope stability monitoring: millimetric precision and multi-tiered averaging to mitigate lag in open cut mining
Slope stability in open cut mining faces mounting pressures from deeper excavations and global mineral demands, elevating risks to personnel and productivity. Recent incidents of slope failures at mining operations in Turkey and North America reveal the consequences of undetected deformations and highlight the imperative for advanced, high-precision monitoring capable of early detection. Global navigation satellite system (GNSS) hardware and software solutions, leveraging integrated software intelligence, deliver millimetric accuracy that surpasses standard real-time kinematic (RTK) centimetre-level performance. Permanently operating to collect raw positioning data every 10 seconds, the system applies an interquartile filtering (IQF) process to generate independent millimetric results in very short periods. This dense dataset averaging enables quicker, reliable insights into subtle slope movements, complementing line-of-sight technologies like radar or InSAR, where environmental obstructions may limit coverage and the 3D vector of displacement is not inherently available.
ADVANTAGES OF PERMANENTLY POWERED GLOBAL NAVIGATION SATELLITE SYSTEM WITH HIGH-FREQUENCY EPOCHS IN MINING
GNSS monitoring systems are often designed to minimise power consumption by keeping the devices in ‘sleep mode’ for most of the time and then intermittently waking up to collect a burst of a small number of readings which are averaged to create a single measurement.
The paucity of the data available to these systems results in low precision when compared to permanently powered devices that provide continuous data collection at a typical rate of one reading every10 seconds. Advances in battery/solar technology now mean that, in many open pit mining environments these permanently operating systems, can be deployed as easily as their intermittently powered alternatives, and this brings the core advantage of dense datasets that enable faster, millimetric results.
In open cut mining, where slopes can deform suddenly due to blasting, rainfall, or structural weaknesses, this continuous high-frequency sampling provides a granular view of movements, reducing the time to achieve reliable precision from days to short periods. Traditional methods like periodic surveys or less frequent GNSS logging often miss transient events, but continuous operation with 10-second epochs captures thousands of data points rapidly, allowing swift averaging and outlier rejection for millimetric outputs. This quicker processing is vital for early detection of potential failures to allow time for verification and mitigation.
Resilience in mining’s harsh environments is another key benefit. Powered continuously, GNSS hardware endures dust, vibrations, and extreme weather without line-of-sight dependencies, outperforming radar in obstructed pits or InSAR during atmospheric interference. Studies on GNSS in coal mines show how dense epoch collection tracks subsidence with high accuracy, minimising downtime by facilitating proactive excavation adjustments. The 10-second rate builds comprehensive time series for multi-tiered analysis, supporting tactical alerts and strategic forecasting. Economically, it lowers costs by reducing manual interventions, with technical reviews noting sub-centimetre repeatability that optimises pit designs. Overall, continuous powering with high-frequency epochs creates a ‘living’ monitoring system, ensuring data integrity for effective risk management in dynamic slopes.
IMPROVING PRECISION THROUGH INTERQUARTILE FILTERING
While RTK GNSS typically achieves centimetre-level accuracy through real-time carrier-phase corrections, the IQF method elevates precision to the millimetric scale via statistical outlier rejection. The hardware, ruggedised for continuous operation in harsh mining conditions, captures high-frequency epochs to build dense datasets. Embedded software employs IQF filtering, computing the first (Q1) and third (Q3) quartiles of the raw coordinate time series and deriving the interquartile range (IQR = Q3 − Q1). Outliers beyond 1.5 × IQR from the median (or up to 3 × IQR in noisy environments) are excluded from the calculated average position. The resulting measured position typically has millimetric precision and is robust against multipath, atmospheric scintillation, or satellite geometry issues. This not only mitigates non-Gaussian noise but preserves the deformation signal, detecting movements as small as 1–2 mm – critical for early creep or acceleration identification in pit walls and benches.
The IQF approach to GNSS data processing has been validated by studies aimed at achieving sub-millimetre repeatability and detecting millimetric scale deformation in other challenging environments, such as landslide monitoring, volcanic deformation and sea level altimetry. The ability of systems using these techniques to outperform intermittently powered devices has also been observed at open pit mines and tailings storage facilities.
TECHNICAL COMPARISON: INDEPENDENT FIXED AVERAGING VERSUS ROLLING AVERAGES
From the dense 10-second epochs, embedded software generates multi-tiered products, highlighting differences between independent fixed averaging and rolling averages. Independent fixed averages compute discrete snapshots at intervals using filtered data: a 5-minute average (60 epochs) reduces noise rapidly, the short-period millimetric IQF offers operational baselines, and a 24-hour average (8,640 epochs) establishes daily cycles. These provide uncorrelated positions with <1 mm deviations, suitable for benchmark reporting but introducing full-interval lags – up to 5 minutes or 24 hours – that may delay acceleration detection in failures.
Rolling averages employ overlapping windows, updating on new fixed inputs for efficiency. A 24-hour rolling average slides over recent 8,640 epochs, 48-hour (17,280 epochs) smooths bi-daily trends, and weekly (60,480 epochs) filters long-term noise. Fixed products avoid error propagation but mask intra-window changes, while rolling averages can halve lag via incremental weighting, boosting responsiveness by 40–60% in nonlinear deformations per time-series analyses. Longer rolling average periods, however, induce averaging lag, damping spikes and postponing alerts by 12–24 hours in rapid failures, as seen when early accelerations were missed.
In slope-monitoring scenarios, fixed 5-minute averages flag blasts with minimal delay, enabling tactical trigger action response plans.
Short-period millimetrics isolate vibrations for creep detection, cutting lag to short intervals. 24-hour fixed confirm trends but risk delays in exponential movements. Rolling 24-hour reduces lag for velocity deltas, improving sensitivity by 20–30%; yet 48-hour/weekly may overlook escalations in displacement caused by, for example, rainfall events.
ADVANTAGES OF HAVING BOTH OUTPUTS AVAILABLE
Access to both fixed and rolling averages from the same sensor enables more nuanced decision-making. Fixed averages deliver stable, independent baselines for regulatory reporting and long-term planning, free from cumulative errors. Rolling averages offer dynamic, low-lag views for immediate interventions, adapting to evolving conditions. This duality minimises false alarms through validation, optimises resources by matching analysis to context – tactical/short/fixed for alerts, strategic/long/rolling for trends – and fosters proactive strategies, ultimately boosting safety and operational efficiency in open cut mines.
AVAILABILITY OF DATA QUALITY METRICS
When expertise is used to analyse and interpret monitoring data, the inclusion of low-quality measurements may sometimes be unavoidable and, therefore, awareness of the presence of these measurements is critical.
Likewise, with the increasing reliance on automated procedures and interpretation using, for example, artificial intelligence (AI) requires the ability to identify and quantify the quality associated with individual measurements to avoid misleading output and AI hallucinations. All monitoring systems suffer periods of lower quality results, but in the case of permanently powered GNSS systems using IQF processing, the quality metrics can be inherent making these datasets suitable for inclusion in automated monitoring systems.
INTEGRATION WITH LINE-OF-SIGHT TECHNOLOGIES
The availability of high-quality GNSS data also opens the door to future advancements in integration and fusion with line-of-sight technologies like radars and InSAR via intelligent software that shares critical data bidirectionally. Precise GNSS positions could calibrate radar phases or resolve InSAR ambiguities, while radar scans feedback to GNSS for enhanced trend validation. Machine-learning-driven platforms would fuse datasets, enabling hybrid alerts where GNSS velocities inform radar thresholds, reducing blind spots and improving failure prediction through interoperable, native software exchanges.