Joint Inversion of Audio-Magnetotelluric and Dual-Frequency Induced Polarization Methods for the Exploration of Pb-Zn Ore Body and Alteration Zone in Inner Mongolia, China

Abstract

Models of subsurface structures are important for successful deposit exploration, but are challenged by the need to integrate data from different geophysical methods. In the present study, we evaluated a method of joint inversion in which audio-magneto telluric (AMT) and dual frequency induced polarization (DFIP) data sets are inverted simultaneously to produce a consistent 2D resistivity model to show a clear image of subsurface structures. To achieve the objectives, we conducted AMT and DFIP surveys along the same survey line within the Dongjun lead–zinc deposit in inner Mongolia by measuring 31 AMT survey sites with a station spacing of 40 m on a 1440 m survey track and operated in fifty-three frequencies in the range of 1–10,400 Hz to record the resistivity distribution of subsurface to depths exceeding 800 m. The same survey setup up was applied to the DFIP method using a pole–dipole array configuration and operating frequencies of 4 Hz and 4/13 Hz. The two-dimensional (2D) model obtained from AMT data revealed distinct low-resistivity anomalies in the middle of the 2D inversion model. In contrast, the DFIP inversion model showed a high resistive body in the same region with relatively high percent frequency effect (PFE) indicating high chargeability. In response to the discrepancies observed in the separate 2D inversion models, we implemented a joint inversion for both the AMT and DFIP data sets. The joint inversion resistivity model shows surficial conducting bodies and a high conductive body along the profile with relatively high PFE, indicating high chargeability. The final joint inversion resistivity model clearly images the large silica alteration zone and the Pb-Zn mineralization. This study demonstrates the feasibility of a joint inversion methodology and highlights the value of integrating geophysical methods through joint inversion for enhanced characterization and exploration of lead–zinc ores.

1. Introduction

In recent years, there has been a notable surge in the advancement of geophysical exploration technology, which has significantly contributed to the field of mineral resource exploration [1,2,3]. The electromagnetic (EM) method and the induced polarization (IP) method are considered the most significant technologies developed for mineral exploration [4]. The development of electromagnetic (EM) geophysical methods in the 1950s has played a crucial role in mapping the lateral and vertical variations in subsurface resistivity. These methods, such as natural source AMT and controlled source (CSAMT), have found a wide range of applications in metallic mineral exploration [5,6,7,8,9,10,11,12], groundwater studies [13], and geothermal system investigations [14,15,16,17]. Induced polarization (IP) has a longstanding history in geophysical applications, initially utilized in mining geophysics for the delineation and localization of ore bodies, as well as in groundwater exploration [18,19,20,21,22,23]. In addition to the existence of advanced geophysical exploration techniques, it is important to note that each method possesses distinct advantages, limitations, and a variety of potential outcomes when applied for mineral exploration [24]. However, when the data from different geophysical techniques are inverted separately, the results can produce inconsistent inversion models; this can lead to difficulties in interpretation and create confusion about the real subsurface structures. The observed discrepancies in the inversion models indicate the potential benefits of employing a joint inversion approach to seek a single, consistent inversion model and reduce the ambiguities in the final model, which satisfies the physical properties of the two geophysical methods.

In a joint inversion methodology, multiple sets of data are simultaneously inverted [7,25,26,27,28]. The existing literature provides numerous examples of methodologies and case studies of joint inversion techniques [25,29,30,31,32,33]. Recently, several geophysical approaches have been merged for joint inversion in order to overcome the shortcomings of individual geophysical prospecting techniques [34]. In this study, we carried out joint inversion of a geophysical data set comprising natural source AMT and controlled source dual-frequency domain induced polarization DFIP data. Initially, individual data sets of AMT and DFIP were inverted separately to produce a 2D inversion model to estimate resistivity structures sensitive to each specific survey. This process involved analyzing the unique characteristics that each method illuminated in the subsurface structures. The resulting models from these independent inversions were then carefully examined to identify both the similarities and differences in the resistivity profiles they revealed. Subsequently, a joint inversion of the AMT and DFIP data sets was conducted, allowing for simultaneous processing and integration of the information from both techniques. Compared to the analysis of single data sets, this approach permits a more comprehensive and accurate representation of the subsurface resistivity structures, utilizing the strengths and mitigating the limitations of each individual method.

This study is structured around three fundamental research objectives: (1) to develop a joint inversion method that incorporates multiple spatially overlapping geophysical AMT and DFIP data sets; (2) to prove that a joint inversion method increases the accuracy of the resulting 2D resistivity and PFE models; and (3) by applying a joint inversion technique to produce clear subsurface resistive and conductive structures with high resolution. Here, we apply and test these methods in a case study of the Dongjun Pb-Zn-Ag deposit in the Central Asian Orogenic Belt. By conducting a comprehensive analysis of the inversion results, including sensitivity analyses and comparisons with existing geological information, we seek to validate the effectiveness of the joint inversion approach of DFIP and AMT and its potential for practical implementation in mineral exploration projects.

2. Regional Geology

The geographical area known as the Great Xing’an Range is characterized by its location across the Siberian, North China, and Pacific tectonic plates (Figure 1a) [35]. It is widely recognized as a complex assemblage of multiple microcontinental blocks, specifically referred to as the Erguna, Xing’an, Songnen, and Jiamusi blocks (Figure 1b) [36]. These blocks are delineated by distinct fault lines: the Tayuan–Xiguitu fault separating the Erguna and Xing’an blocks, the Hegenshan-Heihe fault between the Xing’an and Songnen blocks, and the Mudanjiang fault demarcating the boundary between the Songnen and Jiamusi blocks [35,37]. The Phanerozoic period witnessed significant tectonic changes, including the closure of the Paleo-Asian Ocean, the Mongol-Okhotsk Ocean, and the subduction of the Pacific Ocean. These processes led to the formation of the overall structural framework, which occurred through the northwest to southeast amalgamation of microcontinental blocks. Although there is ongoing debate regarding the specific ages and processes involved in the amalgamation, geological researchers generally agree that the Xing’an block was incorporated into the Erguna block along the Tayuan–Xiguitu fault during the Early Paleozoic era. Additionally, it is widely accepted that the Songnen block fused with the composite block along the Hegenshan–Heihe fault during the late Paleozoic era [35,38]. The Paleo-Asian Ocean underwent its ultimate closure during the late Permian to the Early Triassic period along the Xilamulun–Changchun fault, subsequent to which there was a period of regional extension [39,40]. The magmatic events during the Early-Middle Jurassic period were associated with the closure of the Mongol-Okhotsk Ocean towards the Xing’an Mongolia Orogenic Belt [40,41]. Additionally, the combined effects of the Mongol-Okhotsk Ocean closure and the subduction of the Paleo Pacific Oceanic plate could have played a role in the extensive magmatism and associated mineralization [42,43]. Furthermore, the Jiamusi block is widely regarded as an exotic block that underwent tectonic amalgamation with the Asian continent along the Mudanjiang fault in the Jurassic epoch [44].

The lithologies observed in the northern region of the Great Xing’an Range consist of metamorphic rocks belonging to the Paleoproterozoic Xinghuadukou group [45]. This group encompasses a Precambrian crystalline basement, metamorphic rocks from the Neoproterozoic Jiageda Group, and a Paleozoic cover sequence comprising clastic and carbonate rocks from the Cambrian, Ordovician, Silurian, Devonian, Carboniferous, and Permian periods. Additionally, the region exhibits volcaniclastic rocks and coal-bearing seams from the Jurassic and Cretaceous periods [45,46]. The emplacement of intrusive rocks was predominantly observed during the late Paleozoic and Mesozoic periods, with a minor occurrence during the early Paleozoic era. The primary occurrence of magmatic activity during the early Paleozoic era was concentrated in the Mohe, Tahe, and Duobaoshan regions of Nenjiang County [40]. Basic–ultrabasic rocks were primarily generated during the late Paleozoic era, predominantly emerging at the interfaces of geological blocks. During the late Paleozoic and Mesozoic periods, there was a formation of substantial intermediate felsic intrusive rocks [47]. The emplacement of igneous bodies occurred within a shallow crustal environment, manifesting as midhypabyssal, hypabyssal, and ultra-hypabyssal intrusions composed of felsic and intermediate materials. The primary rock formations consist of dacite porphyry, granite porphyry, quartz porphyry, and quartz monzonite porphyry [48].

Figure 1. (a) Simplified tectonic map showing the main units of central and eastern Asia and (b) sketch tectonic map of the Songnen block (modified from [44,49]).

Figure 1. (a) Simplified tectonic map showing the main units of central and eastern Asia and (b) sketch tectonic map of the Songnen block (modified from [44,49]).

3. Ore Deposit Geology

The Dongjun Pb-Zn-Ag deposit is located in the Hulun Buir area, 20 km northeast of the city of Erguna in the northern part of the Great Xing’an Range. It lies in the eastern segment of the Central Asian Orogenic Belt (50°21′30″–50°23′ N, 120°17′–120°23′ E), which is in the center of the Erguna Block and to the northwest of the Tayuan Xiguitu fault.

The Dongjun deposit comprises several distinct strata, including the Tamulangou formation, the Manketouebo formation, and quaternary sediments (Figure 2). The Tamulangou formation consists of a sequence of volcanic rocks with intermediate to elemental compositions, such as andesite, basaltic andesite, andesitic tuff, sedimentary tuff, volcanic breccia, and small dacite. On the contrary, the Manketouebo formation primarily comprises intermediate to felsic volcanic and volcaniclastic rocks, including rhyolite, volcanic breccia, and volcanic agglomerate. The quaternary Holocene is distributed throughout the river valley. It mostly comprises gray-black humic soil, fine sand, medium sand, pebbles, gravel, and other alluvial materials; humic silt and other swamp deposits; and a residual slope (Figure 2). The region shows several fractures and faults that follow a pattern of NEE, NNW, and N-trending orientations. These geological features are believed to be genetically related to the Genhe fault. The distribution of the orebodies in the Dongjun deposit is controlled by NNW and N-trending faults (Figure 3), which serve as secondary structures of the Genhe fault [46]. The mineralization occurrence in the Dongjun Pb-Zn-Ag deposit is closely linked to late Yanshanian intrusions in the area. The intrusive rocks found in the Dongjun deposit during the Yanshanian period primarily consist of granite porphyry. Granite porphyry and adjacent andesitic tuff, andesite, and sedimentary tuff characterize the Tamulangou formation. These host rocks hold significant importance in the formation.

Mineralization and Alteration

Most host rocks in the Tamulangou formation, such as andesite, andesitic tuff, sedimentary tuff, and granite porphyry, have undergone varying degrees of alteration. The wall-rock alteration can be classified into three zones based on their proximity to the granite porphyry: the potassic–silicic–sericitic alteration zone, the phyllic alteration zone, and the propylitic alteration zone. The potassic–silicic–sericitic alteration zone can be distinguished because it has silicic, alkalic, sericitic, and carbonate alterations. The phyllic alteration zone is characterized by sericitic, silicic, and carbonate alterations. Substantial chlorite, epidote, and carbonate minerals characterize the propylitic alteration zone. The magnitude of alteration typically diminishes as the distance from the underlying granite porphyry increases. The Tamulangou formation exhibits the most extensive mineralization with high levels of silicic alteration [50].

4. Methodology, Data Acquisition, and Data Analysis

The careful selection of appropriate techniques is of utmost importance prior to conducting an extensive geophysical survey plan in a prospective area [51]. To find out if there is mineralization and its geophysical properties, as well as how it relates to different rock formations, geological structures, and mineralization backgrounds, two different geophysical techniques were used to model the subsurface resistivity and to predict the area of mineralization. The key steps of the proposed research are outlined in a flowchart (Figure 4).

4.1. Audio Magnetotelluric (AMT) Method

The magnetotelluric technique for mineral exploration was initially proposed by Andrey Nikolayevich Tikhonov in 1950 [52] and subsequently refined by Louis Cagniard in 1953 [53], and underwent further advancements through the contributions of Cantwell Thomas in 1960 [54,55] and Keeva Vozoff [56]. The method probes into the sub surface’s electrical structure, employing either an artificial (controlled) or natural source electromagnetic (EM) field as the primary source of the field [57,58]. In the case of artificial or controlled magnetotellurics, electromagnetic signals are generated through dedicated EM transmitters. A notable illustration of artificial source magnetotellurics is the radio magnetotelluric (RMT) method, which relies on civilian and military radio transmitters operating within the frequency spectrum of 10 kHz to 1 MHz [59].

Conversely, in natural magnetotelluric exploration, the electromagnetic fields originate from global lightning phenomena, known as sferics (generating short-period signals), and solar wind activities in the ionosphere, which produce long-period signals. Natural magnetotelluric fields can be classified into AMT with frequencies within the range of f = 1–10,000 Hz and broadband magnetotellurics (BBMT) spanning the frequency spectrum of f = 0.001–300 Hz [60]. Fundamentally, the Earth is conceptualized as a horizontal medium, with the magnetotelluric fields representing plane electromagnetic waves projected vertically onto the Earth’s surface [16]. Upon striking the Earth’s surface, a substantial portion of these waves undergo reflection, while only a minor fraction penetrates into the subsurface. This phenomenon is driven by electromagnetic induction, specifically the fluctuating magnetic field, which induces telluric currents to propagate into the subsurface, with the magnitude of these currents being contingent on the electrical conductivity properties [61]. The skin depth (δ), representing the depth within the subsurface where electromagnetic wave amplitude diminishes to 1/e of its value at the surface, is mathematically expressed as follows [62].

$$\delta =\sqrt{{\displaystyle \frac{\rho }{\pi f\mu }}}\approx 500\sqrt{{\displaystyle \frac{\rho }{f}}}$$

(1)

The skin depth is determined by the resistivity (ρ) measured in ohm meters (Ω·m), the frequency (f) expressed in hertz (Hz), and the magnetic permeability (μ) in henry per meter (H/m). It is notable that the primary factors governing the skin depth are the conductivity (reciprocal of resistivity) of geological formations and the operational frequency employed. Geological formations exhibiting enhanced conductivity in the subsurface are commonly associated with materials such as graphite or carbon films, interconnected metallic minerals, aqueous fluids, and partial melt [63].

The above provided expression illustrates that the skin depth exhibits an inverse relationship with frequency. Consequently, lower frequencies possess the ability to penetrate to greater depths, whereas higher frequencies are confined to shallower regions. At the Earth’s surface, orthogonal electromagnetic field components are observed, and these components provide insights into the frequency response, reflecting the distribution of electrical properties within the subsurface medium [64]. The variation in the magnetotelluric field component over time is transformed into a frequency spectrum, enabling the computation of magnetotelluric frequency domain responses, such as apparent resistivity and impedance phases.

The calculation of apparent resistivity can be expressed as follows:

$$\rho ={\displaystyle \frac{1}{5f}}{\left|{\displaystyle \frac{E_x}{H_y}}\right|}^2$$

(2)

In this equation, f represents the frequency in hertz (Hz), ρ signifies the resistivity in ohm meters (Ω·m), E_x denotes the electric field x-component, and H_y represents the magnetic field y-component.

4.2. Dual Frequency Induced Polarization Method

The DFIP method, created by Chinese academician Jishan He [49], represents a significant advancement in geophysical exploration techniques, particularly in the domain of mineral exploration [65]. The (DFIP) method is an advanced geophysical technique operating within the induced polarization (IP) frequency domain. This system integrates both a transmitter and a receiver, essential for its functionality. The core mechanism relies on the utilization of both high-frequency and low-frequency electrical currents. The transmitter, a pivotal component of the DFIP system, is tasked with synthesizing and energizing the electromagnetic field source. This generated field is then strategically deployed into the subsurface geological layers. When this field interacts with subterranean rock formations and ore deposits, it induces polarization effects within these materials [66].

The DFIP system’s receivers are specifically designed to detect these induced polarization responses. The distinctive feature of this methodology is its effectiveness in characterizing the differential frequency characteristics presented by diverse rock and ore types. The principal measurements obtained by this system include the high-frequency potential difference (∆VH), the low-frequency potential difference (∆VL), from which we calculate the resistivity (Ohm-m), and the percent frequency effect (PFE) (

Table 1. Rock and ore specimen electrical resistivity in the study area.

Rock Ore Specimen	Resistivity Range ρ (Ω·m)	PFE (%)
Basalt	1000~2000	0.7%~1.5%
Rhyolite	1500~2800	0.5%~1.2%
Breccia	2000~3000	0.3%~1.0%
Pb-Zn Ore	500~1000	~15.0%

). These parameters are critical in analyzing the subsurface geological structures and identifying potential mineral deposits [66].

4.3. Data Acquisition

In this study, we carried out an AMT survey at the Dongjun deposit area, deploying 31 AMT stations along a 1440-m profile line aligned perpendicular to the geological strike The stations were spaced 40 m apart. We utilized the GSEM-W10 system, developed by Giant Sequoia Artificial Intelligence Technology Co., Ltd. (Hunan, China), for collecting time-variant field or time series data. The AMT data encompassed a frequency range from 1 Hz to 10,400 Hz, covering 53 frequencies. At each station, data collection lasted for 35 min, capturing two horizontal electric field components (E_x and E_y) and two orthogonal magnetic field components (H_x and H_y), shown in Figure 5. The orientations for the X- and Y-directions were north and east, respectively. We measured magnetic field variations using induction coil magnetometers (ICMs) and electric field variations with two pairs of non-polarizable lead–chloride electrodes (Pb-PbCl₂). To lower contact resistivity, each electric field measurement point was pre-saturated with water. We also assessed each station to identify potential sources of interference, such as roads, high-voltage power lines, and communication cables, which were prevalent along the survey lines. The AMT time series field data collected were processed using GSEM-pros software (version 1.0.3). This involved converting the data to the frequency domain and calculating the cross-power spectra. The cross-power spectra calculations were crucial for estimating the impedance tensor, which varies with frequency. This impedance tensor is key for understanding the dimensionality and strike direction of subsurface structures in our study area.

In the same survey line, the DFIP method using a pole–dipole array configuration was conducted. The receiver array, comprising non-polarizable potential electrodes (MN), was arranged at 40 m intervals along a 720 m length. Data acquisition was performed at 31 stations using dual frequencies of 4 Hz and 4/13 Hz, utilizing the SQ-3C model. The measurement process initiated with an initial current electrode (AB) spacing of 80 m (Figure 6). The supply dipole’s center was positioned at the midpoint of the survey line. Subsequently, the distance of the power supply dipole within the survey line was progressively increased by 80 m for each measurement, following a linear sequence from 80 m up to 800 m (in increments of 80, 160, 240, up to 800 m). Beyond the survey line, the distance of the power supply dipole (AB) was extended linearly in steps, starting from 960 m and continuing until it reached the maximum distance of >2500 m.

4.4. Data Analysis

In this section, we analyzed and displayed processed AMT dimensionality and strike estimation analyses using the MTPy software package, an open-source Python package, for the analysis of the MT data [67].

4.4.1. Dimensionality Analysis

Before inversion, we evaluated the dimensionality of the AMT data to determine whether the phases at a given frequency, calculated impedance tensors, and apparent resistivity matched one-dimensional (1D), two-dimensional (2D), or three-dimensional (3D) geoelectrical structures. Because galvanic distortions do not affect the phase tensor, we used it to analyze dimensionality. The parameter β, known as the phase tensor skew, is important for understanding the complexity and dimensions of AMT data related to the underground structure. The phase tensor in a one-dimensional Earth or stratified subsurface has a circular form, signifying a minimal skew angle (β) [68]. The phase tensor has an elliptical distribution in the context of a two-dimensional regional resistivity structure. In the analysis of a two-dimensional regional resistivity structure, the phase tensor exhibits an elliptical configuration. To obtain error-free data, the value of β approaches 0 [69]. The phase tensor in a three-dimensional Earth has substantial β values and is asymmetric. A rapid lateral displacement in the principal axis of the phase tensor indicates the existence of three-dimensional structures. Generally, when β is less than or equal to 5°, it approximates a two-dimensional (2D) structure. In contrast, when β exceeds 5°, it is classified as a three-dimensional (3D) structure. Therefore, this work employs a criterion of |β| ≤ 5° to delineate a 2D structure. Figure 7 shows that the pseudo-section of the phase tensor at different periods helps in determining the dimensionality of the subsurface structure. In the period interval of 10⁻¹ to 10⁻³ (s), the phase tensor ellipses are dominated by a skew angle of |β| ≤ 5°, suggesting a 2D structure (Figure 7). However, in the period interval of 10⁻³ to 10⁻⁴ (s), the phase tensor pseudo section of some stations exhibits asymmetrical ellipses, with a skew angle of approximately ≥5°, suggesting 3D structure. The overall middle portion of the phase tensor pseudo section between b0, b20, b40, b60, b180, b240, b320, b540, and 560, the subsurface shows a skew angle of ≤5°, suggesting a 2D structure. In this study, in most of the AMT data, the β-values are ≤5° and the phase tensors are elliptic. Consequently, we assume that the majority of the 2D structures dominate the study area.

4.4.2. Geoelectric Strike Estimation

Geoelectric strike analysis of the collected data is used to determine the predominant direction of electric current flow in the studied region. The invariants of the impedance tensor (Z) approach were used [70] and the results are presented as rose diagrams as in Figure 8. Since the strike analysis is inherently ambiguous by ±90°, the study area geology was taken into consideration. The strike analysis at different decades and in all periods (Figure 8) shows current flow directions that vary between 32.5° and 162.5°. At shallow depths (10⁻⁵ to 10⁻⁴ s), the strike direction is roughly 162.5°, indicating a predominant NW-SE orientation of subsurface structures. At intermediate depths (10⁻⁴ to 10⁻² s), the strike gradually transitions to 27.5°, and thereafter to 32.5°, signifying movement towards a more NE-SW orientation with increasing depth. The alteration in strike direction with depth indicates various structural regimes or modifications in geological fabric. Shallow NW-SE-trending strikes may signify surface faults or fracture zones, but deeper NE-SW trends could be associated with regional tectonic structures or deeper mineralized zones. The result obtained from the entire time range aligns with regional structures of the area [46].

5. Inversion

5.1. AMT Data

In this research, we conducted a two-dimensional inversion of AMT data using the Occam inversion algorithm, using the Zondmt2D software package [71]. The Static shift, an artifact stemming from the near surface in homogeneities, has the potential to introduce significant complexities and misleading interpretations of apparent resistivity curves. Such complexities could lead to inaccuracies in delineating geoelectric structures. Therefore, we corrected for static shift by using manual adjustments of curve levels while referencing adjacent curves.

For AMT inversion, several critical parameters and procedures were selected to ensure the precision of our results. We initiated the inversion process by selecting a mesh with a height value of 5 and applying an incremental factor of 1.05, which systematically adjusted model parameters in subsequent iterations. An initial half-space resistivity value of 250 Ω·m, smoothing factor of 1, depth smoothing of 1, and smoothness ratio of 0.5 was selected to optimize the inversion process based on conditions specified in reference material [71]. Additionally, we imposed common model limits, setting a minimum resistivity of 10 Ω·m and a maximum resistivity of 10,000 Ω·m to bound the range of resistivity values within the model. The inversion process was executed iteratively, initially spanning 20 iterations and subsequently undergoing 10 more iterations with parameter adjustments aimed at improving the root mean square (RMS) error. The RMS error, quantifying the difference between the modeled and measured data sets, was assessed as a percentage, with a stopping criterion of RMS errors below 10% considered acceptable for our two-dimensional inverse modeling.

5.2. DFIP Data

A two-dimensional inversion of DFIP data utilizing the Occam inversion algorithm was conducted using Zondres2D software. We initiated the inversion process by selecting a mesh with a height value of 5 and an incremental factor of 1.05, which systematically adjusted the model parameters in subsequent iterations. An initial half-space resistivity value of 300 Ω·m was established as a reference point for subsurface resistivity. The parameters, including a smoothing factor of 0.01, depth smoothing of 1, a smoothness ratio of 1, and a focused threshold of 0.05, were each selected to optimize the inversion process based on the conditions specified in reference [72].

The inversion results showed a root mean square (RMS) error for the resistivity model, and the PFE inversion model is about 6.2% after 10 iterations, reflecting a high level of accuracy in our results.

5.3. Joint Inversion Data

We used a joint inversion method to combine data sets from DFIP and AMT surveys, using ZondRes2D software [72]. In the joint inversion method, the process began by selecting crucial parameters, like the minimum and maximum resistivity range (10 Ω·m–22 kΩ·m), the number of layers (50), and an incremental factor of 1.05 to iteratively adjust the model parameters for modification and stopping criteria, either the number of iterations (10) or an RMS error of 0.1. To establish a baseline for subsurface resistivity, an initial half-space resistivity value of 500 Ω·mwas set. Moreover, specific optimization parameters were carefully chosen, including a smoothing factor of 0.01, depth smoothing of 1, a smoothness ratio of 1, and a focused threshold of 0.05, tailored to enhance the inversion process. In this research, we used the Cross-Gradient method and the Gauss–Newton method to effectively integrate data from the DFIP and AMT surveys within Zondres2D [73]. An important aspect of this phase involved adjusting the weight assigned to the MT data within the software. This weight was finely tuned within a range from 0.25 to 1 to improve the configuration between data phases and reduce the misfit, thereby enhancing the accuracy of the inversion results. Additionally, constraints were applied between model parameters to ensure stability and geophysical realistic models. The Occam inversion regularization technique was utilized to promote model smoothness and prevent overfitting, facilitating the creation of realistic geological structures. Following these steps, model parameters were optimized iteratively to create subsurface models that integrated information from both data sets effectively. The RMS error was calculated, aiming for a target error threshold below 10%. Upon completion of the joint inversion analysis, the RMS error for the joint inversion model was determined to be approximately 4.3%, indicating the effectiveness and quality of the generated subsurface model.

5.4. Inversion Results

5.4.1. AMT Result

The resistivity model derived from AMT data over the Dongjun lead–zinc deposit reveals distinct features characterized by low electrical resistivity. The inversion model identifies three zones with moderate to high conductivity (resistivity < 600 Ω·m), labeled as C1, and C2, and two zones with high resistivity (resistivity > 3500 Ω·m), designated as R1 and R2. Notably, the C1 zone is predominantly situated in the surface layers, exhibiting a horizontal extension towards the NE, and could be attributed to the quaternary alluvial deposits of clays and silt. In contrast, the R1 and R2 zones are located in the SW and NE sections of the inversion model and could be attributed to the basaltic rocks. These resistive zones are notably interspersed by the conductive zone C2. In the middle portion of the inversion model, the conductive body C2 is prominent and vertically continuous, and could be related to the lead–zinc ore.

5.4.2. DFIP Result

The resistivity model derived from the DFIP survey conducted over the same area shows numerous unique characteristics. The resistivity model showed one highly conductive body (resistivity < 500 Ω·m) labeled as C1 and two moderate-to-highly resistive bodies (resistivity < 5000 Ω·m) as R1 and R2. The C1 conductive layer is present in the surficial layer, extends in the NE direction, and could be attributed to the quaternary alluvial deposits of clays and silt. The moderately resistive body, R1, is present in the SW section of the inversion model, and could be related to sandy soil gravel or conglomerates, while the highly resistive body, R2, is present in the middle of the inversion model, which could be attributed to basaltic rocks like Rhyolite and Breccia. In contrast with the AMT inversion results, the DFIP shows the presence of a high-resistivity body, R2, in the central part of the survey, where the AMT result identified it as a low-resistivity zone. However, the percent frequency effect (PFE) of the same anomaly shows values greater than 12 percent points, indicating the IP anomaly which could be attributed to disseminated lead–zinc ore within a basaltic rock host.

5.4.3. Joint Inversion Result

The resistivity model derived from joint inversion showed a range of subsurface anomalies that differ from the individual models. These range from highly resistive to highly conductive structures, encompassing a highly conductive layer (C1), a moderately resistive body (R1), and a structure that has high resistivity (R2) at the center of the profile, which overlies a conductive body (C2). The (C1), highly conductive, is present in the surficial portion of the inversion model and has a resistivity value of less 150 Ω·m, which could be attributed to alluvial deposits of clays and silt, while (R1) is located in the SW section, and its resistivity range is less than 2000 Ω·m, meaning it may be related to sandy gravel or conglomerates. In the upper-middle segment of the inversion model, a prominently high resistivity structure, designated as (R2), exhibiting a resistivity exceeding 6000 Ω·m at the center of the profile, is attributed to volcanic rocks such as Rhyolite and Breccia. Situated beneath this high-resistivity zone is a moderately to highly conductive feature, labeled as (C2), characterized by resistivity values below 700 Ω·m, and displaying a notable vertical continuation, and could be attributed to disseminated lead–zinc deposits within silicate host. Additionally, the PFE value associated with this region is greater than 14 percent, which confirms the presence of the IP anomaly related to lead–zinc mineralization.

Figure 9 displays the observed data and calculated resistivity data for all three inversion models (AMT, DFIP, and joint inversion) as obtained directly from the software interface. To further determine the fitting error, cross plots were constructed, as shown in Figure 10, which illustrates the relationship between the observed data and the calculated resistivity data for each inversion model. The fitting errors for the AMT, DFIP, and joint inversion models are represented by the dispersion of points around the 45-degree regression line. Notably, the AMT model exhibits a fitting error of approximately 1.6%, the DFIP model shows 1.5%, and the joint inversion model excels with the lowest fitting error at 1.3%. This plot highlights the advantage of the joint inversion model in accurately representing the subsurface geological structures compared to the individual AMT and DFIP models.

6. Discussion

The AMT inversion model (Figure 11) showed several areas of high- and low-resistivity zones along the profile from SW to NE. In the SW section of the profile, the strata from the shallow to greater depths are observed to be highly resistive, in the range between 800 and 4000 Ω·m. The high-resistivity structure (R1) extends along 400 m (from −500 to −100) along the horizontal distance and is located at a depth of 400 m (from 700 to 300) from the surface. After studying the geology of the area and the resistivity signature, we could interpret that the resistive (R1) zone (3000 Ω·m) could be attributed to the basaltic rocks, like rhyolite, breccia tuff, andesitic basalt, volcanic basalt, and breccia, belonging to the Manketouebo and Tamulangou formations.

The central section of the profile has three distinct anomalous features, which composed of several units of highly conductive zones: (C1) in shallow depths and moderate conductive zones (C2) in the middle of the profile and extending downward. The high conductivity zone (C1) has a resistivity <150 Ω·m and extends 500 m horizontally (from 0 to 500) and 100 m vertically (from 700 to 600) with a northeast orientation. Based on their resistivity signature and the geology of the area, we interpret the (C1) layer to be quaternary deposits of silt, and alluvial clay. The medium electrical resistivity of the conductive body, C2, (<600 Ω·m) is restricted by 400 m of horizontal distance (from −100 to 300) and a thickness of 300 m (from 600 to 300). We deduce that this medium resistivity anomaly could be due to the mineralization of lead and zinc base metals. The central section shows a possible fault: F1, (Figure 11). According to Unsworth [74], faults can increase the permeability parallel to the fault line and inhibit the movement of materials perpendicular to the fault plane. In addition, faults can trap fluids in gouge- and fault-breccia. In this work, the conductivity-enhancing effects of fault F1 are evident in the AMT resistivity model.

The NE section of the profile consists of a high-resistivity structure (R2) with a resistivity of >3000 Ω·m, and strikes southeast, and is 150 m (350 to −500) in thickness and 300 m (600 to 300) in thickness. We interpret that the highly resistive structures (R1 and R2) could be attributed to rhyolite, breccia tuff, andesitic basalt, volcanic basalt, and breccia of the Manketouebo and Tamulangou formations based on geology and resistivity signatures.

In the DFIP inversion model shown in (Figure 12a), a resistive zone (R1) and conductive layer (C1) are observed at a shallow depth. The highly resistive zone (R1) of <2000 Ω·m is located along a horizontal distance of 50 m (from −400 to −270) and located at a depth of 200 m from the surface. Based on the resistivity signature and geological context, this resistive zone (R1) could be interpreted as being basaltic rock, such as andesitic basalt, volcanic basalt, or volcanic breccia. The highly conductive layer (C1) with a resistivity of <500 Ω·m, extending over a horizontal distance of 650 m (from −250 to 400) along the SW-NE section of the inversion model, is attributed to the presence of quaternary alluvial clay and silt deposits, based on the geology and resistivity signature. However, the main contradiction between the resistivity models of DFIP and AMT is in the middle of the profile. In the middle of the DFIP inversion model, a highly resistive body (R2) with a resistivity of >5000 Ω·m is observed, along a horizontal distance of 450 m (from −150 to 300), and extending with depth (from 650 m to 350 m), which is different from the AMT inversion model results as well as the electrical measurement results of rock and ore specimens from the study area (Table 1). In reference to [46] the Dongjun lead–zinc deposit is composed of rhyolite, basalt, andesitic basalt, and volcanic tuff, and exhibits alteration characterized by high levels of silicification. Considering the geology, the literature review, and the high-resistivity signature, it can be inferred that the high-resistivity anomaly (R2) observed in the DFIP inversion model, unlike the low-resistivity anomaly (C2) observed in AMT, is attributed to disseminated mineralization within a silica-rich host, which creates a resistive barrier that increase the resistivity in DFIP, but at the same time increases the chargeability (PFE) values. The IP anomaly zone is located in the middle of the profile (Figure 12b), with a maximum PFE value of 12%. This suggests that the IP anomaly zones are associated with lead–zinc mineralization. The resistivity model derived from the joint inversion of AMT and DFIP data sets (Figure 13a) improve the subsurface anomalies and resolved the mineralization zone that was not clearly identified by AMT or DFIP individually. The joint inversion resistivity model in (Figure 13a) shows a surficial conducting layer (C1), a highly resistive zone (R1), and a moderately conductive structure (C2) along the profile. A conductive layer (C1) which has (<150 Ω·m) is evident in the surficial depth of 50 m thickness from the surface with a horizontal extension of 600 m (from −100–500), and extends in the NE direction. We could interpret the conductive layer (C1) to be due to quaternary alluvial clay and mudstone deposits, based on their resistivity signature and in reference to the geology of the area. The (R1) resistive anomaly is embedded at the upper layers and extends to about 60 m depth from the surface, and has a horizontal extension of 100 m (from no −400 to −300), which could be interpreted as breccia, volcanic basalt, or andesitic basalt. The highly resistive zones (R2) which have (>6000 Ω·m) are located just below the surficial conductive body (C1). The highly resistive body (R2) is located along 450 m (from −280 to 220) along the horizontal distance and at a thickness of 200 m (650–450). In reference to the geology of the area (Figure 2) resistivity signature, we could assume the highly resistive zone (R2) to be basaltic rocks, such as rhyolite, andesitic basalt, or breccia tuff of the of Manketouebo and Tamulangou formations. Below the resistive alteration zone, there is a large area of moderate-to-high conductivity (C2), an anomaly recovered in the joint inversion model, which has a resistivity range (<700 Ω·m) and is distributed across 750 m (from −450 to 300) along the horizontal distance and is located at a depth of about 450 m from the surface, and extends with depth.

The low-resistivity anomaly recovered from the joint inversion model coincides with the known location of the lead-zinc mineralization in the study area (Figure 14a). In addition to a high conductivity signature, there is a high percent frequency effect (PFE) of the anomalous zones shown in (Figure 14b), which confirms the location of the IP anomalous zone. Based on all these inversion models, along with the geological model (Figure 14a,b) and the ore specimen’s electrical resistivity (Table 1), we interpret that the conductive body (C2) to be caused by deep disseminated metal lead–zinc. This anomaly is better resolved by the joint inversion technique than by AMT or DFIP individually.

7. Conclusions

The mineralization in the study area is known to occur within areas of silicification, which is a resistive geophysical target. The two individual geophysical surveys, AMT and DFIP, over the Dongjun deposit imaged the resistivity signature differently due to the nature and physics of each method. To resolve this issue and to comprehensively understand of the subsurface anomalies, we performed a joint inversion using both the AMT and DFIP data sets. The joint inversion resistivity model showed a zone of moderate to high conductivity (C2) extending along the profile with depth. The final joint inversion resistivity model clearly images the large silica alteration zone and area of disseminated Pb-Zn sulfide mineralization based on a low-resistivity signature and high-percent frequency effect PFE anomaly. Further analysis from this joint inversion model highlights potential areas of sulfide and other mineralizations within the silica alteration zone in the form of small or large conductive anomalies. Collectively, this study illustrates that the integration of geophysical methods through joint inversion is a valuable approach for the improved characterization and refinement of exploration-targeting Pb-Zn mineralization zones in future studies.