BlueCap Australia
QTFM Gen 2 sensor AGL height-response field experiment
A four-site BlueCapBird Light field experiment quantifies the response of its integrated QuSpin QTFM Gen 2 sensor from 25 to 50 m sensor AGL around a physical 35 m survey reference.
BlueCap airborne geophysics research
This study is part of the BlueCap scientific experiments archive, where field experiments and published research results are collected as they become available.
Abstract
This article-format research page reports processed condition-level results from a controlled vertical profiling experiment conducted at four agricultural-field sites near Bậc Ray Một and Phước Bình, Xã Bác Ái Tây, Khánh Hòa, Vietnam, on 11 March 2026. The airborne instrument was BlueCapBird Light, BlueCap's suspended airborne geophysical magnetometer system, with a QuSpin QTFM Gen 2 sensor integrated inside it. The experiment tests one deliberately narrow question: how does its azimuth-balanced total-field observation at a nominal hover site change as physical sensor AGL increases from 25 to 50 m?
The experiment produced 204 processed site-height observations. Andrew Musinov conducted the Vietnam field acquisition and piloted the drone; Eugene Podgorbuntsev and Roman Gornov received the acquired results for geophysical processing and did not participate onsite. The study supports acquisition and QC decisions for drone magnetic surveys but does not treat a vertical point profile as proof of continuation error on a spatial survey grid.
1. Research objective and operational hypothesis
Magnetic observation geometry is part of the measurement. Increasing the distance between sensor and geological sources changes the recorded field before line levelling, gridding, derivatives or inversion begin. The operational hypothesis is therefore that physical sensor-AGL control provides a more comparable primary dataset than allowing a broad height envelope and relying on a universal scalar correction.
For BlueCap valley surveys with trees, 35 m sensor AGL is the lower operational and safety reference. The full one-metre compliance corridor is 34.5–35.5 m. Higher acquisition may be necessary for obstacles or safety, but it represents a different observation geometry and must remain visible in metadata and QC.
2. Materials and methods: how the experiment was run
2.1 Field setting and reproducible coordinates
2.2 Instrument configuration and acquisition controls
The analytical dataset contains 204 processed site-height observations. It is a condition-level dataset, not a release of raw 60 Hz, individual-turn, IMU or base-station time series.
System boundary. This experiment characterises the height response of the complete installed BlueCapBird system and its proprietary multi-channel processing workflow, not of an isolated laboratory QTFM sensor. The curves are not raw magnetometer traces. They are final condition-level estimates: orientation-dependent and temporal components are modelled, while motion, navigation, LiDAR and platform channels are retained for selection and QC before the height response is calculated.
3. Results: observed processed height response
Systematic in four ascending profiles — but not one universal correction
Move the selector through the 51 measured height conditions. Each profile was acquired once from lower to higher sensor AGL, so agreement in the direction of change is not a repeatability test. Values are differences from each site's processed total field at the nominal 35 m reference; the curves are observations, not outputs of the 1/r³ benchmark.
- Site A0.000 nT
- Site B0.000 nT
- Site C0.000 nT
- Site D0.000 nT
| Site | Gradient near 35 m | 25–50 m span | AGL RMSE |
|---|---|---|---|
| Site A | -1.2091 nT/m | 29.5175 nT | 0.0804 m |
| Site B | -0.7828 nT/m | 22.0273 nT | 0.0792 m |
| Site C | -1.4171 nT/m | 36.9874 nT | 0.0804 m |
| Site D | -0.9152 nT/m | 26.9153 nT | 0.0886 m |
Curves are processed site-height condition means, not raw 60 Hz series. The 1/r³ benchmark is evaluated separately and is not presented as a universal law.
At 25 m, the processed field is 15.8175–26.7259 nT above the 35 m value. At 50 m, it is 6.2098–10.2615 nT below the 35 m value. The complete 25–50 m response spans 22.0273 nT at Site B and 36.9874 nT at Site C.
Near 35 m, the local gradient ranges from −0.7828 to −1.4171 nT/m. Achieved AGL RMSE against the nominal levels is 0.079–0.089 m. These results quantify site-specific response and do not justify one universal scalar height correction for every geological setting.
4. Discussion: operational meaning of the one-metre corridor
The graph card reports ≤0.672 nT from the processed means at the nominal 34.5 and 35.5 m levels. Separately, after linear interpolation to actual LiDAR AGL, the maximum absolute departure within the physical 34.5–35.5 m corridor is 0.8 nT at Site C; the four site values range from 0.4 to 0.8 nT. These are distinct calculations, not conflicting values. At a one-metre height departure, the local response is approximately 0.8–1.4 nT; larger departures become visibly nonlinear.
An out-of-corridor sample is not automatically unusable, but its comparability depends more strongly on height modelling and local geology. BlueCap applies this evidence in magnetic survey planning and live QC and true sensor-height measurement with LiDAR.
4.1 Commercial relevance for exploration programmes
The commercial value is avoided field and interpretation risk, not the abstract nT number alone. Live corridor QC lets a crew identify and repeat a non-compliant section before demobilisation, reducing the chance of a later remobilisation and expensive re-flight. More uniform physical observation geometry also reduces the processing effort needed to separate flight-path effects from geology and limits unnecessary dependence on regularised continuation. This protects confidence in maps and inversions used to rank drilling priorities; the present point experiment does not itself quantify drill-target displacement.
5. Interpretation boundary and next experiment
The experiment establishes the first operational link: AGL deviation → changed magnetic observation. It does not by itself measure anomaly broadening along a line, prove continuation equivalence, calculate 2D/3D inversion displacement or quantify drill-target error. Those questions require paired coincident horizontal lines or grids physically acquired at multiple heights and compared with upward/downward continuation to a real 35 m reference.
The 1/r³ relationship was not imposed during sensor processing or condition estimation. It was evaluated only after the processed condition estimates had been finalised. The nominal 1/r³ series is therefore a diagnostic benchmark only; its close match is not independent evidence of a universal attenuation law for distributed natural fields.
The 204 condition means do not quantify intrinsic or installed noise and are not a MagArrow II comparison. A defensible instrument comparison requires raw coincident series, common bandwidth, PSD/ASD or fourth-difference noise, turn and line repeatability, crossover statistics and the same physical sensor AGL. Published standalone specifications can provide context, but they cannot replace that field test.
Data and publication status
The corrected processed workbook, reproducible figure code and publication artwork are openly archived in Zenodo dataset 10.5281/zenodo.22054325. The non-peer-reviewed manuscript is published as EarthArXiv preprint 10.31223/X5T789; corrected Version 2 is publicly available at the same DOI. The journal manuscript has been submitted to the Vietnam Journal of Earth Sciences (manuscript 25106).
Project enquiry
BlueCap Starts with the Survey Decision
We use the Portal to turn your survey polygon into a quote-ready BlueCap project, review DEM-aware 2D and 3D mission design, follow field acquisition, and keep processing outputs and deliverables in one client workspace.
No software subscription fee for BlueCap service projects.
