The UAV used for this survey was a commercially available DJI Inspire 2, modified with 3D printed mount to integrate a magnetometer for geophysical measurements (Figure 2). This model was selected for its reliability and advanced features, including redundant key systems, ensuring a high level of safety and multiple camera possibility. It is also equipped with two independent batteries, which allows for efficient battery changes during field operations without interrupting the system or shutting down preplanned flight profiles. This capability allowed for seamless integration of the magnetometer system while maintaining stable flight characteristics essential for high-resolution magnetic surveys (Foss et al., 2025).

In addition to the magnetometer, the UAV was equipped with two cameras. One camera (Zenmuse X4S) was able to closely monitor the movement and status of the magnetometer and sensor during flight, while the second camera (Inspire 2 built-in) continuously monitored the terrain in front of the UAV. The video feed was used only for real time monitoring. Survey planning and execution were managed through the Ground Station Pro interface. The flight time of the UAV setup was approximately 20 min, depending on wind intensity. Since the survey area was in a remote location, batteries were recharged using a petrol generator and managed via the TB50 battery station. With seven spare battery kits and a continuous recharging system, the team was able to conduct virtually nonstop flights with only brief landings for battery replacement.

The magnetometer used for mapping purposes was a compact fluxgate magnetometer, developed at the Czech Technical University in Prague specifically for our UAV magnetometry setup with suspended sensor. Compared to a rigidly mounted sensor, a suspended setup at a sufficient distance of sensor beneath the aircraft has the advantage of eliminating electromagnetic interference generated by the UAV (Accomando et al., 2021; Walter et al., 2021). The cable length of 2.5 m was selected based on system integration tests, which showed that at this distance, interference from the UAV is effectively suppressed below the magnetometer’s intrinsic noise level (see Table 1). The sensor employed two dual-axis ring-core fluxgate sensors with dimensions of 26 × 26 × 6 mm and a flat amorphous core design, as described by Petrucha (2016). The complete device underwent scalar calibration using the method described by Merayo et al. (2000), as well as specific vectorial calibration following the procedure outlined by Janosek et al. (2019). The total weight of the suspended magnetometer system, excluding the battery, is approximately 236 g, comprising 85 g for the fluxgate sensor, 31 g for the GPS module, and 120 g for the control electronics. The full specifications of our setup are listed in Table 1.

The survey area was selected based on the position of the central magnetic anomaly identified in previous airborne magnetic surveys by Geoscience Australia, which mapped a prominent dipolar anomaly in this region (Figure 3). The entire area lies within a flat salt plain, allowing the UAV to maintain a constant altitude relative to the nominal takeoff level throughout the survey. Vertical accuracy of the UAV is approx. 0.5 m (Table 1). Ground Station Pro was used to control the flight. The UAV was programmed to autonomously follow a grid pattern consisting of profiles 3.5 km in length, spaced 25 m apart, and flown at an altitude of 25 m above the ground (sensor altitude 22.5 m ± 0.5 m). This combination of altitude and line spacing was selected to balance lateral resolution, sensitivity to shallow magnetic sources, and the operational efficiency of the UAV platform. Similar parameters have been successfully applied in previous UAV-based magnetic surveys over geologically complex terrains, (Foss et al., 2025; Parshin et al., 2018; Malehmir et al., 2017; Accomando et al., 2023). The UAV flight paths, shown in Figure 3, illustrate the grid coverage superimposed on earlier 200 m line spacing magnetic maps.

To optimize efficiency given the limited battery life of the UAV, flights were planned so that each profile was completed in full before a battery change was required. This ensured that data acquisition was uninterrupted along individual profiles. All flights, excluding landing, were executed autonomously, with the UAV maintaining a direct connection to the base station for real-time monitoring. The operator retained the ability to intervene if necessary. Each flight began at a base station located outside the survey area. Before takeoff, the magnetometer was activated, and data recording was initiated. The UAV ascended vertically to 25 m AGL, moved to the starting point of the profile, and flew along the grid at a constant altitude. At the end of each profile, the UAV turned 90°, moved to the next profile line, and proceeded in the opposite direction. After two consecutive profiles, the UAV returned to the base station for battery replacement. During each battery change, the status of the magnetometer and data recording system was checked to ensure consistent operation. The magnetometer recorded data at a frequency of 62.5 Hz, while the UAV maintained a constant flight speed of ∼14 m/s. For each measurement point, the system recorded three-axis magnetic field data, GPS coordinates, altitude, time, and other sensor parameters. The high-density grid of profiles provided one measured point approximately every 22.4 cm along the flight path and 25 m perpendicular to it. Using this methodology, the survey covered an area of approximately 8 km², with a total profile length of ∼322.72 km. This high-density data acquisition provided detailed spatial coverage of the central magnetic anomaly, enabling the generation of high-resolution maps.

The magnetic data collected during the UAV survey were processed following standard geophysical procedures to ensure data accuracy and reliability (Reeves, 2005). Data points recorded during takeoff, landing, and transitions to the survey grid were excluded to remove measurements outside the survey area. Outliers were identified and removed. Diurnal variation was corrected following Reeves (2005), and heading error was corrected using the directional-averaging method of Hood and Teskey (1989) and Reeves (2005). Since the bias reverses between reciprocal flight lines, correction cancels the offset while preserving geological signal (Pilkington and Keating, 2009). Residual short wavelength line noise (Walter et al., 2019) was then suppressed using standard micro levelling (Minty, 1991) and low-pass filtering (Blakely, 1995). To remove the regional field, the International Geomagnetic Reference Field (IGRF) (Alken et al., 2021) was subtracted from the measured total-field data. The geomagnetic field at the time and location of the survey was approximately 58,000 nT, with an inclination of −64.4° and a declination of 6.2°. By subtracting the measured total magnetic intensity from IGRF data, the residual magnetic intensity (RMI) was calculated. Raw and processed data profiles are shown in Figure 4.

The residual magnetic intensity (RMI) map generated from the UAV data provides a comprehensive view of the central magnetic anomaly. Based on previous surveys, the anomaly was characterized by a dipolar feature with a distinct high and low magnetic intensity zone. However, the UAV data reveal complex shape and additional small-scale variations and edges within the anomaly that were unresolved in earlier datasets. An overlay with the lower-resolution RMI map from previous aeromagnetic surveys (Figure 5B) demonstrates the improved detail provided by the UAV survey.

The Reduction to the Pole (RTP) transformation ideally correct data for the effects of geomagnetic inclination and declination. However, the effect of this transformation is relatively minor in our dataset (Figure 6A), partially due to the moderate local geomagnetic inclination (−64.4° at Acraman), which naturally produces more vertically oriented anomalies, and the influence of remanent magnetization, which RTP is not designed to correct (Baranov and Naudy, 1964; Ugalde et al., 2007). In the survey region, many anomalies remain asymmetrical or dipolar even after RTP, suggesting the presence of remanent magnetization or inclined source geometries. This interpretation is further supported by the Analytic Signal map (Figure 6B), where several anomalies that are dipolar or asymmetrical in RTP appear as symmetrical, high-amplitude features, consistent with remanent effects or complex source geometries. Additionally, regions that appear as negative anomalies on the TMI map often correspond to positive features on the AS map, a characteristic that can indicate remanent magnetization with a direction opposite to the present geomagnetic field (Blakely, 1995; Reid, 1980).

Figure 7 presents the Vertical Derivative and Tilt Derivative maps, which provide further insights into the structure and distribution of magnetic anomalies within the surveyed area. The Vertical Derivative map (Figure 7A) emphasizes shallow magnetic sources by enhancing short-wavelength components, effectively delineating sharp boundaries and near-surface anomalies. These high-gradient zones are concentrated in regions with the strongest RMI amplitudes. The closely spaced positive and negative gradients in these areas indicate rapid lateral changes in magnetic properties, consistent with highly disrupted and multi-directionally magnetized materials likely reflecting remnant of intense fracturing, shock brecciation, and localized melt rock formation in the impact-modified zone.

The Tilt Derivative map (Figure 7B) reveals a more chaotic distribution of magnetic gradients, capturing both shallow and deeper structures that are less pronounced in the Vertical Derivative map. Unlike VD, which primarily attenuates deeper signals and amplifies near-surface sources, TDR normalizes amplitude variations, making it sensitive to a broader range of sources. In our case, this results in a complex, irregular pattern without clear systematic alignment, consistent with the extensive fracturing and disruption expected in the central uplift. Elongated anomalies are present and are consistent with large positive and negative areas in RMI, but the overall distribution is disordered, likely reflecting the intense deformation and multi-phase remanent magnetization associated with the Acraman impact event.

Euler Deconvolution yielded depth solutions for both structural indexes SI = 1 and SI = 2, providing insights into the depth distribution of magnetized sources (Figure 8). The depth values represent the estimated position of the upper edge of the magnetic body. For SI = 1, depths ranged from 3.5 m to 100 m, with a mean depth of 38 m. For SI = 2, the range toward deeper values (20–102 m) with a mean depth of 70 m, indicating a potential shift towards detecting deeper structures. SI = 1 returned approximately twice as many valid solutions consistent with a greater number of shallow features being captured under this assumption compared to SI = 2.

Following standard procedure in Oasis Montaj, Euler solutions were retained only when algebraic misfit was ≤15%, which constrains the formal 95% confidence interval of each solution to ±0.15H (Pilkington and Keating, 2009; Thompson, 1982). For the mean depths retrieved here, this equates to ±6 m for SI = 1 and ±10 m for SI = 2. These values represent the expected individual error bounds, and are consistent with the resolution expected from a 10 × 10-cell window over 10 m data grid.

The spatial distribution of depths also differs by structural index. SI = 1 emphasizes a dense array of shallow anomalies between 10 and 30 m across much of the survey area. SI = 2 highlights more clustered, deeper anomalies. In both cases, coherent depth gradients and localized clusters suggest the presence of vertically zoned source bodies.

2D plots (Figure 8) illustrate the variations in solution distribution and depth between the two SI values, while a 3D visualization with 5× vertical exaggeration provides enhanced spatial context (Figure 9). The 3D surface of top of magnetic source was gridded from inverted 3D Euler solutions for SI1 and was overlaid with the RTP RMI plot, allowing direct correlation between the magnetic anomaly and subsurface structure. These results refine the understanding of the Acraman Crater’s central magnetic anomaly by offering multiple depth models based on different structural assumptions.

The RAPS analysis (Figure 10) identified magnetic source up to depths ∼90 m. These results closely align with the depth estimates derived from Euler Deconvolution, which also indicate major magnetic sources within the 30–100 m range. RAPS was calculated for regirded area of RTP RMI as shown in Figure 9. Three linear segments in and yield slopes of −1.142 ± 0.068 rad (−65.4° ± 3.9°), −0.672 ± 0.057 rad (−38.5° ± 3.3°), and −0.332 ± 0.034 rad (−19.0° ± 2.0°), corresponding to mean depths of 26 ± 3 m, 54 ± 5 m, and 91 ± 6 m, respectively (95% confidence). These values follow standard spectral depth interpretation (Spector and Grant, 1970).

The (Figure 6A) shows many anomalies remain asymmetrical or dipolar after the RTP transformation, suggesting the presence of remanent magnetization or dip of the source bodies. This is consistent with findings that impact melt and highly shocked lithologies can retain multi-directional remanent magnetization not fully corrected by RTP alone (Pohl et al., 1988; Pilkington and Grieve, 1992).

The AS map (Figure 6B) enhances the detection of magnetized bodies irrespective of polarity, revealing multiple, spatially distinct magnetized regions. Notably, anomalies that appear dipolar or asymmetrical in the RTP map manifest as symmetrical, high amplitude features in the AS map, indicating remanent effects or complex source geometries. Furthermore, regions that are negative in the TMI or RMI maps often correspond to strong positive anomalies in the AS map, reflecting remanently magnetized sources with magnetization directions differing significantly from the present-day geomagnetic field (Blakely, 1995; Reid, 1980).

The VD map (Figure 7A) emphasizes short-wavelength features, delineating sharp lateral boundaries of shallow magnetic sources. These high-gradient zones likely correspond to regions of intense fracturing and localized brecciation, consistent with mechanical disruption in the central uplift. The correlation between high VD gradients and AS intensity peaks reinforces the interpretation of near-surface magnetized material. Depth estimates from Euler deconvolution and Radially Averaged Power Spectrum analyses indicate that many of these features are shallow, with sources at depths of 10–30 m (Figures 8, 9), supporting the presence of shock-fractured basement or fragmented impact melt (Williams et al., 1996).

The TDR map (Figure 7B) captures a more chaotic distribution of magnetic gradients, including deeper structures not fully resolved in the VD map. This complexity reflects the multi-phase and heterogeneous nature of the central uplift, where intense deformation and variable remanent magnetization produce intricate magnetic signatures (Pilkington and Grieve, 1992; Grieve et al., 1991). The lack of systematic alignment in the TDR map likely results from impact-induced brecciation and fracture networks, leading to disordered, multi-scale magnetic sources.

The Acraman impact event significantly altered the pre-existing volcanic lithology, producing a complex central magnetic anomaly. Similar features in other large impact structures, such as Sudbury (Pilkington and Grieve, 1992), Manicouagan (Eitel et al., 2016), and Chicxulub (Rebolledo-Vieyra et al., 2010), have been attributed to the remanent magnetization of impact melt sheets and highly shocked basement lithologies. These structures often exhibit multi-phase remanent magnetization, reflecting intense fracturing, localized melting, and variable cooling of impact-modified rocks.

Outcrop measurements within the Acraman structure display an order-of-magnitude variation in magnetic susceptibility (χ) (Williams et al., 1996). Although χ records only the induced component of magnetisation (Dunlop and Özdemir, 1997), such pronounced lateral variability indicates sharp contrasts in magnetic-mineral assemblage created by patchy melting, shock heating and post-impact hydrothermal alteration. Laboratory and field studies have shown that where magnetic mineralogy is heterogeneous, equally heterogeneous remanent components (TRM/CRM) are commonly generated (Pilkington and Grieve, 1992; Henkel and Reimold, 1998). Thus the observed χ variation is not a direct measure of remanence, but it is fully consistent with a lithological framework in which spatially variable remanent magnetisation can develop within the central uplift.

Depth estimates from Euler deconvolution (Figure 8) and Radially Averaged Power Spectrum analyses (Figure 10) provide additional context for the vertical (depth) distribution of magnetic sources within the central uplift. The Euler analysis identified shallow magnetized sources at depths ranging from 5 to 100 m, with SI1 highlighting shallow features (5–30 m) and SI2 capturing deeper structures (20–100 m). The RAPS analysis supports this interpretation, revealing multi-layered magnetic sources with well-defined depth distributions, including shallow (∼26 m), intermediate (∼53 m), and deep (∼90 m) components. It is important to note that in Oasis montaj, the window size used in Euler deconvolution (10 × 10) directly influences the maximum depth of solutions. A larger window size allows for the detection of deeper sources, while a smaller window size is more sensitive to shallow features. Therefore, the maximum depth estimates obtained are inherently limited by the chosen window size during analysis.

This multi-layered structure suggests that the central uplift contains a heterogeneous assemblage of fragmented, variably magnetized blocks, consistent with the intense mechanical disruption and variable cooling rates expected in impact-modified terranes. The observed correlation between shallow sources and high AS intensities further supports this interpretation, as these regions likely host shock-fractured basement or fragmented melt rocks.

UAV magnetometry has proven to be valuable tool for refining the geological and geophysical understanding of impact structures. The complex magnetic pattern observed within the Acraman central anomaly suggests a history of shock metamorphism, partial melting, and possible hydrothermal alteration, which are common in large impact structures (Williams and Schmidt, 2021; Gilder et al., 2018). Previous studies have noted that deeply eroded impact structures often retain magnetic signatures that provide insights into subsurface processes (Rebolledo-Vieyra et al., 2010; Eitel et al., 2016). The elliptical magnetic low and high-amplitude dipolar anomaly identified in previous aeromagnetic data (Williams and Gostin, 2010) align well with the UAV-derived results, confirming that remanent magnetization is a key factor influencing the observed anomalies.

One of the most significant implications of these findings is the potential for targeted future drilling in the central uplift region. Previous studies have proposed that deep drilling into the ‘hot shock’ zone could recover zircons with U–Pb ratios reset by the impact, providing a more precise age for the event (Williams and Schmidt, 2021). The UAV survey results, which delineate the spatial distribution of the magnetic anomaly offer valuable guidance for selecting optimal drilling locations. The present 8 km² UAV grid was intentionally centred on the strongest magnetic high of the Acraman central uplift, providing the resolution needed to refine this key anomaly while acknowledging that it captures only a small portion of the 85–90 km structure. To place these results in a crater-wide context, future work could include radial UAV transects that link the central grid to the annular ring fault, together with targeted palaeomagnetic and rock sampling to constrain magnetization directions. A multidisciplinary extension incorporating high resolution Bouguer gravity, shallow seismic imaging, or near surface EM soundings would further reduce model non-uniqueness. Bruno et al. (2024) demonstrated that such integrated datasets can significantly improve structural interpretations in complex tectonic settings. Applying a similar strategy at Acraman could therefore enhance both depth control and lithological discrimination across the entire crater.

In addition to its scientific advantages, the UAV-based survey offers a highly cost-effective alternative to traditional airborne or handheld magnetometry. The survey covered an area of 8 km², with a total of 322 km of profiles. The UAV operated using two batteries simultaneously, requiring 50 battery pair replacements throughout the survey. To maintain continuous measurements, we used 12 batteries (6 pairs) in rotation, ensuring uninterrupted operation by constantly recharging in cycles. The total operational battery cost for this survey was approximately $90 (considering price of the battery and cycle lifespan). Recharging consumed 1 kW over 24 h. These practical considerations highlight the established advantages of UAV magnetometry in remote or logistically challenging terrains, such as the Acraman Crater. While these benefits are well documented in the literature (Cunningham et al., 2018; Walter et al., 2020; Archer, 2023), our results further demonstrate how UAV systems can enable dense spatial sampling, flexible deployment, and the potential for repeat surveys in previously inaccessible environments.