Assessing future exploration potential of the central and southern Cobar District using integrated 3D geological modelling and geophysical inversion of magnetic data

3D Inversion of magnetic total gradient data in the presence of remanent magnetization

Selection criteria and feasibility of the inversion model for azimuthal electromagnetic logging while drilling (LWD)

SUMMARY Marine magnetotelluric (MT) and controlled source electromagnetic (CSEM) methods have been routinely applied to survey the crustal and upper mantle electrical resistivity structures beneath the sea floor. In practice, there are inevitably site gaps and contamination by noises in the collected data because of lost ocean bottom electromagnetic receivers, unusable data, and difficulties in deploying instruments near deep trenches. So far, it remains unclear to what degree those factors will lower the resolution and the credibility of marine MT and CSEM inversion models. In this paper, we investigate the individual and combined effects of site gaps and data noises on the inversion models through synthetic analyses based on a simple block resistivity model and a realistic resistivity structure derived from the Mariana Trench. The results suggest that data with a sufficiently high signal-to-noise ratio can reasonably recover the subseafloor structures in the area of data gap. The transverse electric mode and tipper data from the MT method are much more sensitive to the structure near the site gap. The joint inversion of MT and CSEM data would improve the model's resolution at the site gap area. The inversion of data with a relatively low signal-to-noise ratio, for example, 10 per cent, can recover the structures with few artefacts if there is no site gap. But if the site gap and noisy data are combined, even a joint inversion cannot correctly recover the burial depths and geometries of the anomalous bodies beneath the site gap where vertical strips are likely present. To improve the model's resolution and suppress inversion artefacts, we propose constraining part of the model with as much a prior information as possible. Specifically, for a survey in the subduction zone, we could reduce the penalties on the model's smoothness at the upper and low interfaces of the resistive subduction slab, or even fix the resistivity of the resistive slab with the help of other information, if any. The inversion models shown in this paper provide valuable references for the site design before marine MT and CSEM surveys as well as for interpreting real data inversion models that may be subject to the same biases introduced by the site gap and noise.

High‐resolution Sea Beam bathymetry and sea surface magnetic data have been collected between 25°S and 27°30′S on the Mid‐Atlantic Ridge. Analysis of the morphology and structure of the ridge axis and flanking terrain as well as the results of a three‐dimensional inversion of sea surface magnetic data are presented. This 227‐km‐long portion of the Mid‐Atlantic Ridge is offset 42 km by the right‐lateral Rio Grande Transform at 25°40′S and 9.5 km by the left‐lateral nontransform Moore Discontinuity at 26°30′S. A third nontransform discontinuity, the Midway Discontinuity, exists near 26°12′S and is defined by a small displacement of the axial magnetization distribution and the disruption in the along‐strike continuity of ridge‐parallel terrain elements. Although the detailed survey presented in this paper extends out to only 5‐ to 7‐m.y.‐old crust, a regional compilation of magnetic data from this area by Cande et al. (1988) indicates that the relative positions and dimensions of the spreading cells have characterized this part of the Mid‐Atlantic Ridge for at least the past 55 m.y. The offsets dimensions and geometries, however, have changed markedly because of prolonged differential asymmetric spreading between segments. Between and within the accretionary units that are spreading at uniform rates (35 mm/yr total opening rate), the morphology of the ridge axis varies dramatically, from a well‐defined rift valley (20–25 km wide, 1–1.5 km deep) to an axial swell that rises 1500 m above the adjacent valley floor. The axial magnetic structure is also highly variable along‐strike. The inversion solution shows a series of generally linear, short‐wavelength (<10 km) magnetization highs within the central Brunhes anomaly over each of the four segments in the survey area. These anomalies are interpreted to be the central anomaly magnetization high that is observed over the neovolcanic zones of many of the world's mid‐ocean spreading centers. The distinct axial magnetization highs can extend out to 2‐m.y.‐old crust at ridge‐discontinuity intersections, suggesting that in some instances they may also be the product temporal variations in the geochemistry of the extrusive rocks, e.g., highly fractionated and strongly magnetized basalts enriched in iron and titanium. The magnetic inversion shows that the central anomaly is locally attenuated at the minimum depth points of three of the four spreading segments in the survey area. Both inversion and forward modeling of magnetic data show that topography alone cannot produce the anomalously low values observed. In the case of an axial swell centered at 26°S, attenuation of the central anomaly is attributed to the effective thinning of the source layer by demagnetization effects of a crustal thermal anomaly and/or a zone of substantial alteration. Taken together, the bathymetric data and magnetic modeling document strong along‐axis gradients in crustal and upper mantle properties.

Inversion of magnetic data has been an important tool for the interpretation of the data. The conventional inversion approaches cannot recover geologically acceptable models in the presence of remanence. Hence, some techniques have been developed for the inversion of magnetic amplitude data which is independent of magnetization direction. The conventional methods in the inversion of magnetic amplitude data give rise to blurred models with long tails. In this study, a new algorithm has been developed in favor of sparse inversion of the data that generates more focused models using the Cauchy norm. The algorithm applies a regularized conjugate gradient (RCG) method to acquire the inverse model. The new algorithm has been employed for the inversion of two data sets from synthetic examples and magnetic data sets over the Allahabad iron deposit in Iran and the Osborne copper–gold deposit in Australia. All data sets exhibit remanent magnetization. The results showed that the new algorithm generates robust solutions that are geologically acceptable.

The inversion of magnetic data is inherently non- unique with respect to the distance between the source and observation locations. This manifests itself as an ambiguity in the source depth when surface data are inverted and as an ambiguity in the distance between the source and boreholes if borehole data are inverted. Joint inversion of surface and borehole data can help to reduce this nonuniqueness. To achieve this, we develop an algorithm for inverting data sets that have arbitrary observation locations in boreholes and above the surface. The algorithm depends upon weighting functions that counteract the geometric decay of magnetic kernels with distance from the observer. We apply these weighting functions to the inversion of three-component magnetic data collected in boreholes and then to the joint inversion of surface and borehole data. Both synthetic and field data sets are used to illustrate the new inversion algorithm. When borehole data are inverted directly, three-component data are far more useful in constructing good susceptibility models than are single-component data. However, either can be used effectively in a joint inversion with surface data to produce models that are superior to those obtained by inversion of surface data alone.

The opening of the Scotia Arc resulted in the final breakup of the land bridge between South America and the Antarctic Peninsula. The South Orkney Microcontinent (SOM) constituted part of this former connection and it is now the largest continental block in the Southern Scotia Arc. We present the first 3D model of the SOM that, given its strategic position and characteristics, allows us to advance the knowledge of the tectonic processes involved in the development of the Scotia Arc. Due to the scarcity of reliable geological data, the initial approximation of the deep structure of the SOM was supported by the calculation of three main geological boundaries from geophysical data: the acoustic basement, the boundary of the magnetic anomaly source and the Moho depth. The 3D model was built, refined and validated by forward modeling and joint inversion of gravity and magnetic data. We have accurately defined the geometry of the sedimentary cover, determined the geometry of the intrusive igneous body causing the Pacific Margin Anomaly (PMA) and mapped the heterogeneity of the crustal thickness. These structural features show a clear relationship to each other and are consistent with an important E‐W extension to the east of the SOM during early stages of the Scotia Arc formation, prior to the opening of the Powell Basin.

Summary The paper proposes the approach to 3D modeling and 3D inversion in which the observed data are the second and third invariants of the magnetic gradient tensor. This enables not only to detect the bodies with abnormal magnetic properties but also the magnetization vector direction. This, provides additional opportunities for further geological interpretation. In order to calculate the abnormal magnetic field, we propose to use two approaches: analytical and numerical. The analytical approach is used in the case when the abnormal magnetic bodies are characterized by only remanence and/or magnetic permeability differing from the vacuum magnetic permeability not more than 10%. We present the results of comparison of the numerical and analytical methods and results of 3D inversion on the synthetic magnetic gradiometry data, which are obtained for the complex model containing two bodies with remainence; same direction as the Earth's magnetic field and the distributed structure with remanence in the opposite direction to the Earth's magnetic field. The approach proposed for mathematical processing the magnetic gradiometry data allows not only the general detection of the abnormal magnetic structures but also the delineation of local bodies the magnetization vector direction which is different from the magnetized surrounding structure.

The 11GA-YO1 deep seismic reflection survey reveals information on the crust down to ~66 km depth, imaging the crust-mantle boundary. The seismic survey traverses the Yilgarn Craton, Officer Basin and Musgrave Province yielding information on their key structures and boundaries. Interpretation of the seismic reflection data was complimented with forward modelling and 3D inversion of gravity and magnetic data. This allowed the geological structures interpreted in the seismic data to be investigated and extended into 3D space. The 3D gravity and magnetic inversions shown here reveal information on the geology and structure of the crust in the Yilgarn-Officer-Musgrave (YOM) region. In particular, information on the nature of dipping bodies beneath the Officer Basin and the boundary between the Yilgarn Craton and Musgrave Province.

Regional gravity and magnetic surveys are essential sources of information about the structure and geodynamics of the lithosphere. However, geologically meaningful inversion of gravity and magnetic data usually requires integration with other geophysical methods. We have developed a 3-D joint inversion framework that has the flexibility of using independent inversion codes and model discretizations for each of the included methods, is easily expandable and supports a wide range of the coupling constraints. Here we show its application to the regional geophysical datasets available in Ireland. We present the results of joint inversion of long-period magnetotelluric data, seismic traveltimes, and land gravity – a multiparameter geophysical model of the crust and uppermost mantle of the whole Ireland. On a smaller scale, we present the results of joint inversion of gravity, airborne magnetic and magnetotelluric data for the Limerick Basin, focusing on imaging of a Carboniferous volcanic structure.  The main aim is to better understand the Pb-Zn mineral systems which are controlled by the tectonics of the basement and lower crust. Exploration-scale geophysical surveys and geothermal exploration will also benefit from the regional 3-D geophysical models.

Canadian Malartic is an Archean low-grade bulk tonnage native gold deposit. The deposit is mostly located in altered clastic metasedimentary rocks, mafic–ultramafic dykes, and monzodioritic porphyry intrusions. Airborne magnetic and frequency-domain electromagnetic (EM) data were inverted to reconstruct the geological units associated with the mineralization, especially the intrusive masses. The 3-D inversion of magnetic data, which used a tetrahedral mesh to a depth of 2.4 km, shows that mafic volcanic rocks and iron formation rocks extend to depth in the area, more so than diabase dykes. The magnetic inversion also shows that the diorite and monzodiorite rocks of the Lac Fournière A pluton are dipping toward the south on its northern edge at the contact with the metasedimentary rocks. The 1-D inversion of the frequency-domain EM data, for both electrical conductivity and magnetic susceptibility, is able to reconstruct geological structures to a depth of approximately 100 m, providing more details and information about these features. The intrusive masses such as diabase dykes, diorite and monzodiorite rocks, and mafic volcanic rocks are reconstructed as electrically conductive structures in the inversion results. The metasedimentary rocks are resistive, and the overburden is conductive in most of the area. The geophysical data and inversion results suggest the presence of some features (such as diabase dykes and monzodiorite rocks) that are not yet present on some parts of the geology map. A comparison of the EM-derived susceptibility and the magnetic-derived susceptibility over the iron formations can reveal the effect of remanent magnetization.

Subglacial geology remains largely unknowns in Antarctica. Direct geological samples are limited to ice free regions along the coast, high mountain ranges or isolated nunataks, while the origin of geological material transported by glaciers themselves is often ambiguous. 3D singular and joint inversions of airborne gravity and magnetic data recovers subsurface density and susceptibility distribution. The relationship between both inverted petrophysical quantities provide crucial insight for subglacial geology and rock provinces interpretations. Validation of indirect derived subglacial geology models are critical but very challenging in Antarctica due to the sparsity of rock samples.We present 324 new density and susceptibility measurements on rock samples from Northern Victory Land, East Antarctica. 251 samples have been measured at the National Polar Sample Archive (NAPA) from the Federal Institute for Geosciences and Natural Resources (BGR) in Berlin-Spandau, Germany and additional 73 samples were measured at the BGR in Hannover, Germany. We use the petrophysical measurements to validate our recent regional scale 3D joint inversion model of the Wilkes Subglacial Basin and the Transantarctic Mountains. Furthermore, we validate inversion results on a local scale of singular magnetic inversion based on high resolution airborne magnetic data with a flight line spacing of 500m in the Mesa Range.We demonstrate that we can provide reliable discrimination between Ferrar Dolerites, Kirkpatrick Basalt and Granite Harbour intrusion rocks based on our local inversion model and that the recovered susceptibilities agree with those measured at rock samples from the study area. Furthermore, we show that regional scale inversion model of gravity and susceptibility distribution agrees for samples of the dominant crustal rock types. However, densities of small-scale dense intrusion bodies like Ferrar Dolerites are underestimated by the regional scale inversion, while the susceptibility range is correctly recovered.Constraining subglacial geology with joint inversion of airborne potential field data is heavily depended on the resolution of the airborne survey, flight line coverage, the inversion scale, and the scale of the target feature. Regional scale inversion is adequate for large scale geological heterogeneities, which underestimate petrophysical quantities for small scale structures, while local scale inversions are able to resolve such structures but are more computational demanding and in the case of Antarctica lack ultra-high resolution airborne gravity data with a line spacing below 1000 – 500m.

A method is developed for 2D forward modeling and nonlinear inversion of gravity data. The forward modeling calculates the gravity anomaly caused by a 2D source body with an assumed depth‐dependent density contrast given by a cubic polynomial. The source body is bounded at depth by a smooth, curvilinear surface given by the Fourier series, which represents the basement. The weighted and damped discrete nonlinear inverse method presented here can invert gravity data to infer the geometry of the source body. The use of the Fourier series to define the basement geometry allows the interpreter to reconstruct a broad variety of geometries for the geologic structures using a small number of free parameters. Both modeling and inversion methods are illustrated with examples using field gravity data across the San Jacinto graben in southern California and across the Sayula basin in Jalisco, Mexico. The inversion of the San Jacinto graben residual Bouguer gravity data yields results compatible with those from previous interpretations of the same data set, suggesting that this geologic structure accommodates about 2.5 km of sediments. The inversion of the residual Bouguer gravity data across the Sayula basin suggests a maximum of 1‐km‐thick sedimentary infill.

Serra Geral Group plumbing system and 3D geological framework in Morungava region, southern Paraná–Etendeka LIP (Brazil): Combining 3D geologic and magnetic models

The three-dimension (3D) inversion of magnetic data is an effective method of recovering underground magnetic susceptibility distributions using magnetic anomaly data. The conventional regularization inversion method has good data fitting; however, its inversion model has the problem of a poor model-fitting ability due to a low depth resolution. The 3D inversion method based on deep learning can effectively improve the model-fitting accuracy, but it is difficult to guarantee the data-fitting accuracy of the inversion results. The loss function of traditional deep learning 3D inversion methods usually adopts the metric of the absolute mean squared error (MSE). In order to improve the accuracy of the data fitting, we added a forward-fitting constraint term (FFit) on the basis of the MSE. Meanwhile, in order to further improve the accuracy of the model fitting, we added the Dice coefficient to the loss function. Finally, we proposed a multi-constraint deep learning 3D inversion method based on UNet++. Compared with the traditional single-constraint deep learning method, the multi-constraint deep learning method has better data-fitting and model-fitting effects. Then, we designed corresponding test models and evaluation metrics to test the effectiveness and feasibility of the method, and applied it to the actual aeromagnetic data of a test area in Suqian City, Jiangsu Province.