Methodological foundations for interpreting modern airborne geophysical survey data

Recent decades have seen a systematic increase in the share of airborne geophysical work within geological exploration as a whole. Its volume has also grown in absolute terms, particularly in recent years, owing to a sharp rise in prices for major mineral commodities. Demand for airborne geophysical surveys has increased because their effectiveness and exploration returns have improved sharply compared with similar work in earlier years. This raises an interesting question: what distinguishes modern geophysical surveys from earlier work, and which principal factors account for their significantly greater effectiveness?

In our view, these factors include the introduction of microprocessor technologies into geophysical instrumentation, enabling digital signal recording and processing. This in turn made it possible to improve the noise immunity of the measurement channel substantially and increase instrument sensitivity sharply. Digital signal recording and high-performance onboard computers also increased measurement-channel throughput, markedly improving survey spatial resolution. Nevertheless, the principal factor responsible for a truly revolutionary increase in airborne geophysical survey effectiveness is the use of satellite navigation systems to georeference measurements. Their adoption not only improved horizontal positioning accuracy by more than an order of magnitude, but also enabled highly accurate vertical positioning.

Figure 1 presents the results of a 1:50 000 aeromagnetic survey conducted in one area of Rudny Altai in the mid-1980s and repeated in 2005. The survey area has highly dissected terrain, with an elevation range exceeding 1500 m. Survey techniques were similar in both cases: parallel lines spaced 500 m apart, broadly following the terrain, a recording rate of at least 1 measurement per second, and a magnetometer sensor in an external bird. Nevertheless, the difference in results is evident: the overall field structure is broadly reproduced, but the modern survey has substantially greater resolution, particularly for local targets.

The reason for this difference appears clear: the Doppler positioning system used in the earlier surveys could not provide horizontal positioning accuracy better than ±25 m. Observation-point elevations were not taken into account at all, necessitating considerable data smoothing during levelling.

scale 1:50 000

scale 1:10 000

Fig. 1. Comparison of aeromagnetic survey results.

Interestingly, the formal survey-quality indicators—the root-mean-square error calculated at intersections between regular survey lines and a specially flown diagonal line—differ only slightly. Among other things, this demonstrates that this quantity is not an entirely adequate measure of survey quality.

Thus, the principal differences between modern and earlier surveys are satellite navigation for observation-point positioning and digital signal recording and processing. These survey characteristics should receive primary consideration when determining whether archival materials are suitable for reinterpretation using modern technologies.

Improvements in the quality and resolution of airborne geophysical surveys impose greater demands on data-interpretation technology. The methodological foundations of airborne geophysical data interpretation are, in turn, determined by the characteristics of the acquisition technology and the typical properties of the objects studied, including the following:

1. The rapid nature of airborne geophysical technologies and their ability to cover large areas with high-spatial-resolution surveys in a short time; the ability to operate in inaccessible areas; and the absence of anthropogenic disturbance to the survey areas, allowing investigations where other types of work are prohibited.
2. The three-dimensional nature of the information acquired, reflected in the fact that both the elevation and the height above ground of observation points generally vary during a survey. Airborne gravity surveys are an exception, as they are usually flown at a specified elevation.
3. The feasibility and merits of using a sufficiently comprehensive suite of methods: adding an information channel generally increases the cost of the work by no more than 10–15%.
4. Work is frequently undertaken in poorly studied areas, where a priori geological and geophysical information appropriate to the survey scale is virtually absent or insufficient.
5. The ability to operate at virtually every stage of geological exploration while obtaining data of the required detail and representativeness; high-accuracy horizontal and vertical positioning of observation points; and an additional information channel—a digital terrain model—derived from satellite navigation data and radio-altimeter readings.

These characteristics of airborne geophysical survey technology and of the objects studied allow general requirements to be formulated for interpretation technologies used to analyse airborne geophysical data.

1. Airborne geophysical results generally provide the basis for planning detailed ground geological and geophysical investigations. This requires final information to be obtained within a relatively short time. The time spent on interpretation should therefore not substantially exceed the time required for the survey, so interpretation technologies must be largely automated and standardised.
2. Since airborne geophysical surveys are usually undertaken in poorly studied areas, the interpretation technologies and algorithms employed must remain effective when a priori information is scarce.
3. The shortage of a priori information can be partly compensated for by integrated interpretation of heterogeneous airborne geophysical data. Elements of integrated interpretation should be used both in preliminary analysis and in synthesising the results.
4. The frequent absence of an up-to-date geological base for the study areas requires interpretation procedures addressing both general geological tasks—geological and structural-tectonic mapping, landscape-geological zonation and the like—and targeted tasks aimed at locating prospective objects.

The foregoing establishes the need and rationale for an initial general geophysical interpretation of the data to address standard geological tasks (total interpretation). The subsequent processing workflow should aim to locate exploration targets using direct criteria and indicators (local interpretation) and indirect ones (targeted interpretation). Targeted and local interpretation of the same airborne geophysical data may be repeated many times, focusing on different objects according to the task. In this sense, airborne geophysical surveys are sufficiently versatile, a consideration that should inform the selection of survey methods, particularly at regional stages of work.

The question of what should be regarded as the input airborne geophysical data for interpretation merits separate consideration. The principal source of instability in geophysical inverse problems is interference of various origins in the original information itself. Unfortunately, its effects are still often artificially amplified by interpreting substitutes for the original data—what are commonly called maps. Clearly, all surveys are actually conducted on irregular observation networks, and observation-point elevations may vary widely. Nevertheless, maps of field elements are traditionally constructed before interpretation begins. To do this, formal interpolation, often entirely disregarding observation-point elevations, reduces the original data to flat, horizontal, regular grids. Only after this operation, which introduces additional errors, do the principal analytical and synthetic interpretation procedures begin.

Less commonly, preliminary mapping uses approximation methods to reduce the original data to a horizontal plane at the highest observation-point elevations. This approach is certainly more appropriate, but it too introduces errors into the original data. In effect, it analytically continues the measured field upward to the reduction level, suppressing the high-frequency part of the field recorded at lower survey levels. This leads to the first key interpretation principle:

The input information for interpretation must be electronic digital databases, rather than maps produced by formal interpolation without accounting for observation-point elevations.

Maps can serve only as convenient and familiar, but not entirely adequate, means of visualising these data. Interpretation programs must use all three coordinates of each observation point together with the value of the field being interpreted. Consequently, the resulting models will not contain distortions caused by inadequate treatment of observation-point elevations.

During modelling, no formal transforms of the original fields may be subjected to quantitative interpretation; only source-like approximations may be used for field separation. Interpolation introduces distortions specific to the method employed, which, under conditions of instability, can distort the separated components substantially and almost unpredictably.

This does not mean that we oppose transformations in general: they are crucial for detection and recognition tasks. In our view, however, the anomalous fields themselves should not be transformed. Fields depend on many different factors, including the observation-point elevation differences mentioned above, different source depths and, in magnetic prospecting, interference from conjugate induction extrema, which manifests differently in different regions of the Earth. Traditional recognition approaches involve first formally interpolating the field onto a regular grid and then transforming the interpolated fields. Finally, every transform must be examined directly to avoid mistaking conjugate extrema unrelated to geological structure for manifestations of the objects studied, which is highly inefficient operationally. Auxiliary transformations such as reduction to the pole do not fully resolve this inefficiency. Conjugate extrema remain afterwards, although they are indeed somewhat easier to identify than in the original fields.

Accordingly, the second methodological principle can be formulated as follows:

When separating field components caused by different geological sources, only source-like approximations should be used to construct meaningful parameterised models. Detection and recognition tasks can then be addressed by analysing these models, including the use of formal transformations.

The foregoing applies primarily to potential-field methods, since airborne gamma-ray spectrometric data, for example, are reduced to ground-surface level during initial processing. Nevertheless, visualising these data using conventional planar representations—geophysical maps—can also lead to substantial interpretation errors because it does not account for the association of anomalies with specific elevation levels (Fig. 2).

These considerations must be taken into account when performing substantive interpretation of airborne geophysical data. At the same time, physical-field maps remain the most convenient and familiar means of data visualisation, and abandoning them entirely would apparently be premature, even with the reservations stated above.

We now consider the components and characteristics of general geophysical data interpretation for the principal airborne geophysical methods.

Potential-field methods (gravimetry and magnetometry). The methodology for interpreting magnetic and gravity anomalies is determined primarily by the combined influence of observation-point elevation differences (mainly in aeromagnetic surveying) and terrain relief. This factor has a particularly strong effect on anomalous fields. Figure 3 shows a synthetic example of DT anomalies calculated at aeromagnetic survey points for rocks forming an uneven terrain surface and magnetized uniformly by induction in the direction of the Earth’s normal field. The ground-surface profile (the lower curve in the figure) and the flight path are actual records from a survey in Buryatia.

The combined effect of these factors produces a relatively complex calculated DT field. Geologically, the model medium is completely homogeneous, while the amplitude of the terrain-induced anomalies naturally increases with the magnetization of the terrain-forming rocks and may reach several hundred nT. The morphology of these anomalies is highly complex and, as even this example shows, cannot be reduced to a simple correlation with topography. The character of the magnetic field would not fundamentally change if the survey were flown horizontally at a specified barometric altitude, or equivalently at a specified level above the geoid surface.

Under these conditions, even the simplest qualitative interpretation techniques familiar to geophysicists—visual analysis of observed fields and their moving-window transforms—become entirely meaningless. A human interpreter cannot simultaneously and effectively evaluate so many heterogeneous factors contributing to anomalous fields. Strictly speaking, the transformations themselves can be performed only where the terrain is flat and horizontal. Given the instability inherent in inverse-problem solutions, where even infinitesimal interference or inadequacies in the approach lead to damaging consequences, traditional technologies that disregard observation-point elevation differences become not merely ineffective but harmful.

The most popular way of addressing these difficulties, adopted by many geophysicists, is to first reduce the anomalous field to a horizontal plane at the highest terrain elevation. Various technologies designed for flat, horizontal terrain can then be applied. This approach is nevertheless not optimal: during reduction, local anomalies recorded at low observation-point elevations are effectively analytically continued upwards by hundreds of metres or even kilometres, and are consequently smoothed and obscured to an excessive degree.

Under these complex conditions, we consider three-dimensional modelling aimed at constructing a petrophysical model from observations at their actual locations to be the most effective interpretation approach. The computer automatically accounts for variable observation-point elevations, ground-surface topography and, in magnetic prospecting, conjugate induction extrema arising from nonvertical rock magnetization and unrelated to any geological factors. Once the model has been constructed and these interfering factors eliminated, it becomes feasible to analyse it using transformations, including those designed to recognise geological objects.

In this respect, the optimal model is a subhorizontal layer with strictly lateral variations in density or magnetization, for which uniqueness of the inverse solution was proved by V.M. Novoselitsky in 1965 [1]. Ideally, one would like a model consisting of several such layers that directly approximates complex three-dimensional structures, but direct construction of such models is theoretically prohibited. These so-called grid-type models have a specific deficiency: automated fitting within them inevitably makes the model pathological.

• For potential-field methods, total interpretation within the approach described above can address the following geological tasks:
• Investigating the morphology of subhorizontal interfaces, such as the buried basement surface or individual horizons within the sedimentary cover, using singular-point localisation methods [2];
• Compositional and petrophysical mapping of the area, at both exposed and buried surfaces, using the above-mentioned technology for modelling potential-field data as a layer bounded by two contact surfaces, followed by two-dimensional interactive classification of the resulting effective-density and effective-magnetization distribution models [3, 4];
• Structural-tectonic mapping using specialised lineament-analysis technologies applied to effective-density and effective-magnetization distribution models [4];
• Three-dimensional geological mapping based on assessment of the morphology of anomalous objects using interpretational tomography methods [5].

Airborne gamma-ray spectrometry. Airborne gamma-ray spectrometric (AGS) data essentially reflect the geochemical structure of an area and are primarily aimed at assessing its mineral potential. They also contain information about the structure of the uppermost part of the section, which can be extracted through landscape-geochemical zonation.

Preliminary assessment of the prospectivity of individual poorly studied areas should begin from a systems perspective [6, 7, 8]. This is based on the well-known proposition that large deposits can form only where favourable global, regional and local ore-controlling factors coincide. For assessment of medium- to large-scale AGS data, the hierarchy of subordinate geological-structural levels comprises the following sequence: metallogenic province (n × 10⁴–10⁵ km²)—ore district (n × 10³–10⁴ km²)—ore cluster (n × 10²–10³ km²)—ore field (up to 5 × 10² km²)—deposit (a few km²). The principal characteristic of economically significant ore objects is the superposition (“telescoping”) of favourable criteria at every hierarchical level without exception (Fig. 4). Objects that fail to manifest at even one level are generally of no interest and should not be given substantial weight when delineating prospective areas.

A reliable radiogeochemical base covering the entire study area and constructed from AGS surveys at different scales makes it possible to investigate different levels of spatial association between the hierarchy of ore taxa and the corresponding radiogeochemical anomalies. It also enables so-called “throughgoing” prediction, progressively delineating prospective areas at different levels of organisation of matter. The effectiveness of this approach is demonstrated in the works of A.B. Kazhdan [7, 8], E.V. Plyushchev and V.V. Shatov [9], V.A. Arsenyev [6] and others. To implement it with AGS data, a method has been developed for mapping epigenetic radiogeochemical zonation characteristic of different types of ore-controlling metasomatites. It reduces the influence of interfering landscape-lithological factors and isolates the so-called “secondary” component of the radiometric field (R₂) [10]. Calculation of R₂ is based on disruption, by overprinting processes, of the radioactive-element ratios and interrelationships typical of unaltered rocks in the region. The secondary component is assessed by two parameters: sign (S) and magnitude (M). M reflects the intensity of epigenetic alteration in metasomatic structures of different orders (ore district, ore cluster and ore field), while S reflects lateral changes in the dominant nature of radioactivity—that is, the radiogeochemical zonation used to determine the type of metasomatism. Successive analysis of anomalous R₂ values at different levels of organisation of matter establishes regional radiogeochemical zonation of the corresponding rank and identifies local, potentially ore-bearing sites within prospective zones.

Airborne electromagnetic surveying. In recent years, the airborne electromagnetic methods traditionally used in Russia have been fixed-wing dipole inductive profiling (DIP) and very-low-frequency radio surveying (SDVR). Field experience in 2005, however, showed that dedicated very-low-frequency radio stations currently operate for no more than 1.5–2 hours per day, practically precluding efficient coverage of large areas with this method. In the DIP configuration used, the transmitting dipole is rigidly fixed to the aircraft, while receiver coils are placed in an external bird. The increased transmitting-dipole power makes it possible, in particular, to lower the minimum frequency substantially, and provides higher sensitivity and greater depth of investigation than foreign equivalents. A distinctive feature of this measurement system, however, is that its geometry changes continuously during flight. Consequently, in areas where a horizontally layered model is inadequate, the results can be used only for qualitative interpretation. Nevertheless, individual experiments indicate the possibility of developing an original interpretation technology using techniques unconventional in electromagnetic prospecting.

Experimental work at one of Yakutia’s diamond-bearing ore fields yielded very encouraging results (Fig. 5). The work was conducted in an area of trap formations where detailed ground gravity surveying had been completed. In the figure, black curves show the original fields and green curves show approximations reconstructed from the Fourier spectrum using a specified number of harmonics. Contours of the vertical gradient of the anomalous field (Strakhov’s method) and a vertical section of the total normalised gradient (Berezkin’s method) are also presented.

The figure shows that both methods reliably locate the same features. Airborne electromagnetic surveying can therefore be used effectively, at least for mapping trap basalt cover. Furthermore, the feasibility of locating features in electromagnetic survey data using methods developed for potential fields (in the spectral domain) suggests that filtering-based interpretational tomography methods [5] may also be effective for interpreting DIP electromagnetic data. Airborne electromagnetic data are generally used at the targeted or local interpretation stage.

1.

2.

Fig. 5. Singular-point locations derived from ground gravity survey data (1) and from DIP-A airborne electromagnetic data at a frequency of 520 Hz

Integrated interpretation of data based on different physical principles should synthesize the results of comprehensive interpretation of each method, taking account of the elevation level under analysis. As noted above, such studies are usually performed at the local (targeted) interpretation stage. Examples of targeted data interpretation in exploration for various mineral commodities are discussed in other articles in this issue.

REFERENCES

1. Novoselitsky V.M. On the theory of determining density variations in a horizontal layer from gravity anomalies. Izvestiya AN SSSR. Fizika Zemli. 1965. No. 5. pp. 25–32.

2. Babayants P.S., Blokh Yu.I., Trusov A.A. Investigating crystalline basement surface relief using magnetic survey data. Geofizika. 2003. No. 4. pp. 37–40.

3. Babayants P.S., Blokh Yu.I., Trusov A.A. Investigating crystalline basement structure in platform regions using magnetic and gravity survey data. Geofizika. 2003. No. 6. pp. 55–58.

4. Babayants P.S., Blokh Yu.I., Trusov A.A. Capabilities for structural and lithological mapping using magnetic and gravity survey data in the SIGMA-3D software package. Geofizichesky Vestnik. 2004. No. 3. pp. 11–15.

5. Babayants P.S., Blokh Yu.I., Trusov A.A. Interpretational tomography using gravity and magnetic survey data in the SIGMA-3D software package. Theory and Practice of Geological Interpretation of Gravity, Magnetic and Electric Fields: Proceedings of the 31st Session of the D.G. Uspensky International Seminar. Moscow: United Institute of Physics of the Earth, Russian Academy of Sciences. 2004. p. 11.

6. Arsenyev V.A., Bartenev Yu.V. A method for processing airborne gamma-ray spectrometric survey data to identify signatures of radiogeochemically specialized blocks of different ranks. Geologiya Uranovykh Mestorozhdeniy, issue 83, 1983, pp. 123–127.

7. Kazhdan A.B., Pakhomov V.I. Processing primary empirical data within a systems approach to subsurface studies. Sovetskaya Geologiya, No. 11, 1986, pp. 3–11.

8. Kazhdan A.B. Prediction, Prospecting and Exploration of Uranium Deposits. Moscow: Energoatomizdat, 1983. 189 pp.

9. Plyushchev E.V., Shatov V.V. Geochemistry and Ore Potential of Hydrothermal-Metasomatic Formations. Leningrad: Nedra, 1985. 247 pp.

10. Vorobyev V.P., Zubov E.I., Aslamov Yu.V. Hierarchy of secondary radiogeochemical structures as a reflection of ore-metasomatic systems of different orders. Abstracts of the International Geophysical Conference, 15–18 September 1997, Moscow, 1997, G18.