Modeling of the electrostatic field generated by horizontal current sources in a sea-ice-covered four-layer medium Authors: Xueyao Chen, Zhaolong Sun, Chaoli Jiang, Kui Zhu, Mengxiang Li, Runxiang Jiang, Wentie Yang, Kai Yan, Yuedou Du Published: 2026 ABSTRACT AbstractThe distribution of the underwater electric field around a target in stratified media provides a crucial foundation for research on target detection and localization in marine environments. This paper proposed a four-layer analytical model (air–sea ice–seawater–seabed) to address the issue of electric field propagation in ocean environments involving sea ice. Starting from the Laplace equation for a steady electric field, the repeated image method was employed to introduce image current sources. Combined with the boundary conditions, an analytical solution for the electric field intensity within the seawater layer of the four-layer medium model was derived. Subsequently, a numerical model employing the finite volume method (FVM) was constructed according to the conditions of the experiment and compared with the analytical solutions under the same idealized four-layer conditions. Finally, a scaled-down experiment was designed to compare the measured electric field intensity values with the analytical solutions. The results indicated that the relative error between the measured electric field intensity values and the analytical solutions ranged from 1.17% to a maximum of 11.28%, thus validating the effectiveness of the analytical model and demonstrating its applicability in realistic environments. This work provides the theoretical and technical foundations for detecting underwater targets in the sea-ice-covered marine electromagnetic environment.Keywords: Underwater electric field, Horizontal static current source, Four-layer medium model, The repeated image methodSubject terms: Engineering, Environmental sciences, Ocean sciences Introduction The physical fields of vessels include electric fields, acoustic fields, magnetic fields, and hydrostatic pressure fields, which serve as crucial target signals for the identification and localization of underwater detection systems. At present, a variety of advanced underwater multi-physical field detection and measurement systems have been developed internationally, including the Multi-Influence Range System (MIRS) developed by SAES in Spain, the underwater multi-influence module developed by the Polish Naval Academy, and the TECOTEC Magnetic and Acoustic Measurement Range System1–4. These studies further demonstrate the crucial role of multi-field detection in target recognition. Among these detection modalities, electric field detection offers advantages such as being unaffected by hydrometeorological conditions in the marine environment, having stable and reliable detection performance, and possessing strong identification capability. Consequently, research related to environmental and target electric field characteristics has garnered increasing attention in the field of marine environmental target detection and localization. Furthermore, underwater electric fields can be applied to monitoring the corrosion status of vessel hulls and evaluating the effectiveness of anti-corrosion measures, conducting geophysical surveys to obtain seabed geological structure information, facilitating the exploration of seabed oil and natural gas deposits, and locating sunken vessels to support rescue and salvage operations. These diverse application scenarios validate the practical value of underwater electric fields and highlight their wide-ranging utility in the field of marine engineering. Given that marine medium exhibits good electrical conductivity and corrosiveness, dissimilar metals submerged in it form corrosion currents, thereby generating corrosion-related static electric fields (CRSE)5. For vessels, corrosion currents form a loop through the hull, seawater, and propellers, which in turn gives rise to CRSE6–8. Among the existing modeling methods for CRSE, the finite element method (FEM), boundary element method (BEM), and equivalent source method (ESM) are widely applied9–11. Due to its concise mathematical form and clear physical meaning, the ESM has become an important theoretical basis for constructing CRSE models12–14. This method demonstrates that actual sources with arbitrarily complex distributions can be characterized by the linear superposition of a finite set of equivalent electric dipoles or current sources, as shown in Fig. 1. This significantly simplifies electric field calculations and enables adaptation to different seawater electrolyte environments15,16. Equivalent diagram of corrosion-related static electric field. The equivalent source method (ESM) provides a crucial technical means for the inversion of electromagnetic fields from underwater targets. However, in practical underwater environments, when considering the electrostatic field distribution within underwater targets such as vessels and unmanned underwater vehicles (UUV), the issue of multiple layer media distribution should be taken into account17. Sommerfeld pioneered the investigation into the electric field generated by an electric dipole on conductor surfaces beneath a planar interface18. Building upon this foundation, Hoitham proposed an equivalent dipole pair model for the static electric field of vessels, thereby establishing the cornerstone for research on vessel electric field signatures19. Early research primarily concentrated on two-layer medium models. Bannister extended the theoretical framework towards practical scenarios by solving for the electromagnetic fields of horizontal electric dipoles in a two-layer medium, employing finitely conducting earth-image theory techniques20. Meanwhile, Schäfer demonstrated that the underwater potential is subject to the coupled effects of material polarization, measurement depth, and conductivity21. This finding underscores the necessity of accurately accounting for environmental factors in any predictive model. With the advancement of marine detection requirements, the three-layer medium model of “air-seawater-seabed” in the shallow marine environment has progressively become the research focus due to its closer alignment with realistic environments. Cao and Tian utilized CST simulations to demonstrate that increasing the number of medium layers can extend electromagnetic propagation distances, thereby validating the necessity of the three-layer medium model22,23. Further research has refined this understanding. Wang identified low-frequency and interface effects as critical factors for long-distance transmission and quantified the influences of water depth and medium conductivity24,25. Feng extended investigations to complex seabed topography and its impact on field distribution26,27, while Wu systematically investigated the influence of the seabed medium in the stratified marine environment on the electromagnetic field in the air radiated from the extremely low-frequency horizontal electric dipole located in the seawater28. Significant methodological innovations have emerged for solving the three-layer medium model. These include Huang’s complex image theory with vector potential, Zhang’s conclusion on surface-wave-dominated long-distance propagation, Xu’s near-field approximation formula for ELF signals, Wang’s simplified lateral wave model, and Ren’s field decomposition method (separating secondary and background fields)29–33. Collectively, these approaches provide efficient pathways for analyzing the electric field distribution in the three-layer medium. It is noteworthy that a fundamental similarity exists between the propagation characteristics of extremely low-frequency electromagnetic waves and those of static current sources within a three-layer medium. The propagation behavior of both is predominantly governed by the stratified structure and interfacial effects of the three-layer medium. Key parameters, such as seawater depth and the electrical conductivity of each layer, exhibit a commonality in their modulating effects on the electric field distribution and attenuation patterns. Nevertheless, the three-layer medium model struggles to fully capture the complex stratified structures prevalent in actual marine environments. This limitation has spurred the development of four-layer medium models. Woloszyn proposed an analytical four-layer medium model comprising “air-seawater-seabed-non-conductive” layer, enhancing the accuracy of describing complex seabed structures34. Affected by special temperature-salinity structures, dynamic distribution of ice-water mixtures, and periodic freezing-thawing processes of ice layers, the electromagnetic environment in sea areas with ice layers differs significantly from that in conventional sea areas35. The low temperature of ice layers inhibits the migration rate of ions in seawater, reduces the overall medium conductivity, and intensifies the electrochemical polarization effect at the interface between the vessel hull and seawater36. In addition, the freshwater input from sea ice melting and the deep high-salinity seawater form a vertical salinity gradient, driving the salt-gradient battery effect and generating an electrostatic field dominated by ion transport37. Transient salinity fluctuations caused by ice layer freezing-thawing also induce transient currents, which in turn interfere with the stability of vessel’s electric field. Lofgren and Weeks experimentally investigated the correlation mechanisms among sea ice salinity, ice growth rate, and environmental solute concentration38. Islam introduced deep learning techniques to achieve intelligent estimation of sea ice thickness based on dielectric constant inversion39. However, due to the pronounced heterogeneity and environmental sensitivity of sea ice in terms of its physical structure and chemical composition, it is seldom treated as an independent and critical layer in most current electromagnetic modeling studies of multilayered marine environments, thereby lacking dedicated modeling and systematic analysis. This paper proposes an analytical model of the electric field generated by horizontal static current sources in a four-layer medium (air-sea ice-seawater-seabed), aiming at the accurate modeling of the underwater electric field of vessels in sea areas with a sea ice layer. Firstly, based on Laplace’s equation for steady electric fields, the electric potential and electric field intensity generated by a pair of horizontal static current sources in the seawater layer were derived for a two-layer medium model. The repeated image method was then introduced to satisfy the multilayer interface boundary conditions via successive mirror reflections, leading to an analytical expression for the electric field in the seawater layer. Subsequently, a finite-volume numerical model was established under experimental conditions and compared with the analytical solution under the same idealized four-layer scenario to verify its correctness. Finally, a scaled-down experiment was designed, and the measured electric field value was compared with the analytical results to evaluate the applicability of the analytical model in realistic environments. Methods Current sources or electric dipoles are the basic units for the inversion of underwater electromagnetic fields of targets such as vessels. Previous studies have shown that due to the differences in electrical conductivity between the air layer and seabed layer in the marine environment relative to seawater, the distribution of underwater electric fields in the marine environment is significantly affected by different media. As a special medium, sea ice also often has a non-negligible impact on the distribution of underwater electric fields in polar and other low-temperature sea areas40. Therefore, this paper takes the current source in the four-layer medium of sea-ice-covered areas (such as polar regions) as the basic research unit, as illustrated in Fig. 2, and focuses on analyzing the distribution of its underwater electrostatic field under the influence of the ice layer. Consider a horizontally stratified four-layer medium representing the polar marine environment. The domain is defined within a Cartesian coordinate system where the z-axis denotes depth. The individual layers are characterized by their respective macroscopic electrical conductivities (), which represent the air (), sea ice (), seawater (), and seabed (). The parallel planar interfaces between these domains are located at depths (air-sea ice, namely BC1), (sea ice-seawater, namely BC2), and (seawater-seabed, namely BC3). A horizontal current source pair, constituting a generalized dipole with positive current and negative current separated by a discrete distance 2a, is submerged strictly within the high-conductivity seawater layer (Layer 3) at an operational depth . However, it should be noted that in actual marine environments, sea ice is a complex composite consisting of pure ice, brine pockets, and air bubbles. Due to the intricate internal pore structures and the heterogeneous distribution of brine pockets among ice crystals, the electrical conductivity of the ice layer is non-uniformly distributed, with localized regions of higher conductivity. Such complexity is difficult to faithfully simulate under engineering and laboratory conditions. Therefore, this paper proposes a simplified four-layer medium model, in which the electrical conductivity of each layer is assumed to be homogeneous. Four-layer medium model. General formulation of the multilayer boundary value problem Current sources or electric dipoles serve as the fundamental analytical units for the inversion of underwater electromagnetic fields generated by submerged targets such as vessels. While previous analytical frameworks, notably the foundational four-layer horizontal electric dipole model rigorously established by Wołoszyn, have successfully addressed the “air-seawater-seabed-non-conductive layer” stratification prevalent in shallow coastal waters34, the presence of polar sea ice introduces structurally distinct electromagnetic boundary conditions. Therefore, this paper constructs an adapted four-layer analytical model specifically tailored for polar regions, wherein the sea ice acts as an intermediate finite-thickness layer exerting profound conductive constraints. In the steady-state static regime, the spatial electric field E is irrotational, which means it satisfies: By combining the microscopic form of Ohm’s law () with the continuity equation for steady currents (), the scalar electrical potential within each homogeneous layer must satisfy either the Poisson equation (within the source layer) or the Laplace equation (within source-free regions): where is the Kronecker delta function indicating that the physical source resides exclusively in Layer 3. The absolute uniqueness of the potential distribution throughout the entire multi-layered domain is strictly governed by the Dirichlet and Neumann boundary conditions enforced at each geometric interface : Continuity of the scalar potential ; Continuity of the scalar potential ; Continuity of the normal component of the current density . Continuity of the normal component of the current density . Furthermore, at the uppermost boundary delineating the interface between the air and the sea ice surface (), the practically infinite resistance of the air layer () mathematically enforces a zero-flux Neumann boundary condition:, ensuring no current escapes into the atmosphere. Analytical solution via the successive reflection series expansion To resolve the coupled boundary value problem without resorting to computationally intensive volumetric numerical discretization (e.g. FEM), we employ the method of successive reflection series expansions. While conventional approaches often visualize this through discrete heuristic image geometries, the problem is fundamentally an algebraic boundary matching sequence. When a primary electrostatic potential field, originating in a host medium with conductivity , propagates and impinges upon an impedance boundary with an adjacent medium of conductivity , it inevitably generates a reflected potential field back into the host medium. The relative magnitude of this boundary reflection is rigorously quantified by the electrostatic reflection coefficient , which is defined as: For an isolated positive point current source located at a point within an infinitely unbounded medium of conductivity , the singular primary potential is governed by the standard Green’s function: where . However, confined within the finite vertical boundaries of the “ice-water” waveguide bounded by the seabed, this primary field undergoes infinite successive reflections. The total electrostatic potential observed within the seawater layer () is an absolute superposition of the primary source potential and an infinite convergent series of secondary (reflected) multipole potentials. To simultaneously satisfy the continuous boundary conditions at both and , we must synthesize the total solution by summing the geometrically decaying contributions from these reflective interactions. Assuming the effective thicknesses of the seawater and ice layers are denoted as and respectively, the primary reflection coefficients governing the internal reverberations are (sea ice-seawater) and (seawater- seabed). Because the ice-air boundary perfectly reflects outgoing current (), the ice layer itself acts as a compound dielectric reflector influencing the effective coefficient. The m-th order recursive reflection terms can be systematically generated to fulfill the interface conditions. The scalar potential generated solely by the positive current pole anywhere within the seawater layer assumes the generalized summation form: where represents the Euclidean spatial distance from the observation coordinate to the mathematically equivalent location of the m-th order boundary reflection in the p-th geometric permutation state. The discrete amplitude coefficients are determined by the sequential multiplication of the relevant reflection coefficients ( and ) raised to powers corresponding exactly to the number of interfacial interactions the field component has undergone. By taking the negative spatial gradient of the comprehensive total potential (which intrinsically incorporates the identical summation structure for the negative sink pole ), the exact analytical expressions for the three-dimensional electric field intensity components are directly extracted: Modeling of a four-layer medium using the repeated image method For the four-layer model of air-sea ice-seawater-seabed layer, it is necessary to account for the boundary conditions defined between all layers of the model, including BC1, BC2, and BC3. Assuming the depths of the sea ice and seawater are and respectively, and the distance from the current source to boundary BC3 is h. Given that the normal component of the electric field intensity vector equals zero in boundary condition BC3, in accordance with the repeated image method41, it is necessary to introduce an image current source at in Layer 4, as shown in Fig. 3a. The application of the repeated image method in the four-layer medium model. Given that the electric current source is located within the seawater layer (layer 3), the influence of the two medium layers adjacent to the seawater layer is relatively significant, including layer 2 and layer 4. In the three-layer system composed of layers 2, 3, and 4, it is necessary to introduce an image current source (denoted as current source ②) to satisfy boundary condition BC3, and its position is shown in Fig. 3b. However, the intervention of this image current source will disturb the boundary condition BC2. Consequently, in the next-order reflection process, a pair of image current sources (denoted as current sources ③ and ④) should be introduced at height . The newly introduced current sources will again disturb the boundary condition BC3. Consequently, an additional pair of successive image current sources (denoted as current sources ⑤ and ⑥) should be introduced at height . To simultaneously satisfy both boundary conditions BC2 and BC3 and achieve the required computational accuracy, higher-order image reflections should be further considered. In the four-layer medium structure (layer 1-2-3-4), an image current source should be introduced to satisfy boundary condition BC1, with its configuration illustrated in Fig. 3c,d. The electric field excited by the current source located at is primarily distributed within layer 3, resulting in a non-zero normal component of the electric field intensity at boundary BC1. Therefore, a pair of image current sources should be placed at height to fulfill boundary condition BC1 (denoted as current sources ③ and ④), and a subsequent reflection should be applied at height to satisfy boundary condition BC3 (denoted as current sources ⑤ and ⑥). To simultaneously meet BC1 and BC3 and achieve the required computational accuracy, it is necessary to further consider fourth-order and higher-order image reflections. It is worth noting that the magnitude of the image current remains constant and does not vary with the order of reflection. For the proposed four-layer horizontal current source model, the equations for the electric field intensity components generated in the seawater layer are as follows: where and denote the displacement of the current sources relative to the center of the plane in the Cartesian coordinate system, serves as an index to distinguish the electric field intensity generated by the current source with current (positive) or (negative), serves as an index to distinguish the contribution from the real current source versus the contribution arising from the image current source generated during each reflection step, serves as an index to specify the particular interface whose influence is being accounted for within a given reflection step. represents the number of reflections, which is primarily determined by the electrical conductivity of the layered medium and the relative position of the current source within the layered structure. Furthermore, it is worth noting that in most ice-covered sea areas, the current source is located far from the seawater–seabed interface. In such environments, the contributions of image current sources at that interface are weak enough to be negligible. Under such conditions, the results calculated by the four-layer model constructed in this paper are essentially identical to those obtained from the three-layer model consisting of air, sea ice and seawater. However, the presence of the seabed layer becomes significant in shallow water regions. Experimental platform Experiment environment To validate the correctness of the analytical formulas derived in Methods and to specifically investigate the characteristics of the near-region electric field generated by a pair of horizontal current sources, a systematic model experiment was conducted in a low-temperature laboratory. The experimental setup was designed to simulate a four-layer medium environment consisting of “air–sea ice–seawater–seabed”. A fiberglass tank (dimensions: 3.2 m × 2 m × 1.6 m) was employed to reduce external electromagnetic interference and provide structural support for the stratified media. Both the electric current sources and the three-component electric field sensors were mounted on a low-dielectric, low-loss epoxy resin bracket to minimize the influence of auxiliary structures on the electromagnetic field distribution. The initial depth of seawater was set to 1.39 m, and the ambient temperature was maintained at − 15 °C. The initial water temperature was 18 °C, which gradually decreased during the cooling process and eventually stabilized at − 2.9 °C. The freezing process consisted of two stages: it took 5 days from the completion of water filling until a 5 cm ice layer formed on the surface, and an additional 2 days for the ice to thicken from 5 cm to 8 cm. Throughout the freezing process, the electrical conductivity was not constant. To accurately determine the conductivity values, measurements were taken at multiple locations in the tank after the water temperature had stabilized, and the arithmetic mean was adopted as the final value. The resulting measurements indicated a sea ice conductivity of 0.02 S/m and a seawater conductivity of 2.2 S/m. The experimental platform is shown in Fig. 4. The block diagram of the measurement circuit used and the geometric description of the electrode configuration used are shown in Figs. 5 and 6, and the main instruments and equipment used are summarized in Table 1. Experimental platform. The block diagram of the measurement circuit. The geometric description of the electrode configuration. Table 1Instruments used in the experiments.InstrumentsProduct modelParametersActual usage quantityElectric current sourceTitanium oxide electrode–11Bipolar power supplyKIKUSUI PBZ40-10Output voltage: 0–40 VOutput current: 0–10 A1Three-component electric field sensorMiniature three-axis Ag/AgCl electric field sensorSelf-noise≤30nV/rtHz@1HzFrequency response range: 0.01–1080 Hz12Storage recorderMR600016 differential channel3Multi-parameter online water quality monitoring instrumentMQPC-10Electrical conductivity measurement range: 0–20 S/m1 Instruments used in the experiments. Output voltage: 0–40 V Output current: 0–10 A Self-noise≤30nV/rtHz@1Hz Frequency response range: 0.01–1080 Hz Prior to the commencement of the experiment, the three-component electric field sensor was connected and calibrated with the data acquisition unit, and the sampling frequency was set to 500 Hz to ensure adequate capture of dynamic variations in the electric field signals. During the formal experiment, the paired electric current source was connected to a bipolar power supply. To accurately separate the effective signal from environmental interference, background noise was first acquired continuously for 30 s under no-excitation conditions to assess the ambient and system inherent noise level. Subsequently, the bipolar power supply was activated to inject a steady direct current signal into the seawater medium, and the electric field response was recorded for no less than 30 s. Throughout the process, all experimental parameters and ambient conditions were meticulously documented to ensure data reproducibility and result reliability. Equivalence scheme In a traditional experimental configuration, the electric current source is typically held stationary, while the measurement process is accomplished by moving the electric field sensor. However, in the present experiment, the imperative to preserve the structural integrity of the ice sheet was a primary consideration. To prevent potential damage to the ice caused by sensor movement, and based on the principle of equivalence, the strategy of increasing the number of electric current sources was adopted to equivalently substitute for the physical translocation of the sensor. The experimental system features a fixed epoxy resin bracket as its core supporting component. Leveraging the exceptional insulating properties and mechanical stability of the epoxy resin material, this bracket ensures the spatial positions of both the current sources and the sensors remain constant throughout the measurement process, thereby effectively mitigating system errors introduced by mechanical movement. Eleven horizontal electric current sources and twelve three-component electric field sensors were pre-deployed in the experimental setup. Through a pairwise combination strategy of adjacent current sources, an equivalent measurement configuration was achieved, effectively emulating a measurement system comprising a pair of current sources and twelve survey lines. Each survey line corresponds to a fixed spatial path. By sequentially switching the current source combinations along each line, ten discrete measurement points were established per line, thereby enabling systematic acquisition of the spatial distribution of electric field intensity within the seawater layer. This approach replaces the conventional practice of physically relocating electric field sensors, significantly enhancing operational feasibility and ensuring data reliability under sea-ice-covered conditions. The spatial layout of the experimental system is detailed in Fig. 7, and an equivalent schematic diagram is provided in Fig. 8. In the figures, the green spheres represent the preplaced horizontal electric current sources, while the blue and red spheres denote three-component electric field sensors. The distinction in color is used to clearly indicate sensors located in different spatial planes, thereby intuitively illustrating the three-dimensional layout characteristics of the measurement system. The experiment comprised four measurement planes, totaling 12 survey lines. These were designated from top to bottom—starting closest to the ice layer—as Plane 1 to Plane 4. Each plane contained three survey lines, labeled from front to back as Line 1 to Line 3. Eleven electric current sources were aligned horizontally and numbered sequentially from left to right as ① to ⑪. In Fig. 8, the ten measurement points along a single survey line correspond—from right to left—to the equivalent measurement points under the excitation of current source pairs ①–②, ②–③, …, and ⑩–⑪, respectively. This design integrates structural fixity and measurement equivalence, significantly enhancing the accuracy, stability, and operational feasibility of electric field measurements in ice-covered environments. Arrangement diagram of electric current sources and electric field sensors. Equivalent diagram of the layout of electric current sources and electric field sensors. Data process The data acquired by the three-component electric field sensor comprises two distinct phases: the first phase consists of background potential difference measurements recorded over 30 s while the DC excitation source was inactive, primarily representing ambient and system inherent noise; the second phase involves mixed potential difference signals, also lasting no less than 30 s after the DC excitation source was activated, which include both the target electric field response and superimposed background interference, as shown in the raw signal in Fig. 9. To accurately extract the effective potential difference signal generated by the DC excitation source, the average values of the potential difference data over the two 30-s intervals were first computed to suppress random noise and highlight the steady-state response. Subsequently, the difference between these two averages was taken. This differential operation eliminated common-mode background interference, thereby isolating the net change in potential difference attributable solely to the source excitation. The net potential difference value was then divided by the electrode spacing in the corresponding measurement direction to obtain the three orthogonal components of the electric field strength () at the measurement point. Finally, based on the electric field vector magnitude formula , the total electric field strength magnitude at each location was calculated. This value serves as the empirical basis for subsequent comparison and validation with theoretical results. Sample of original data. Results Four cases were designed in the experiment, corresponding to two sets of different parameter combinations (ice thickness: 5 cm, 8 cm; DC source intensity: 30 mA, 50 mA). The parameters of four cases are listed in Table 2, where “Distance1” represents the distance from the electric current source to the lower surface of the ice layer, and “Distance2” to “Distance5” represent the distances from the first layer to the fourth layer of measurement planes to the lower surface of the ice layer. Table 2Case parameters.CaseIce thickness (cm)DC source intensities (mA)Distance1 (cm)Distance2 (cm)Distance3 (cm)Distance4 (cm)Distance5 (cm)Case153036.511.521.541.551.5Case 255036.511.521.541.551.5Case 383033.58.518.538.548.5Case 485033.58.518.538.548.5 Case parameters. When the electric field sensor is positioned near the seabed, where the conductivity is lower than that of seawater, the number of reflections impacts the solution accuracy. For the given parameters (the depth of seawater and media conductivities and ), the number of reflections should be incrementally increased within the iterative computation until the error tolerance criterion is satisfied. The dimensionless components of the electric field intensity were computed for according to Eqs. (21)-(24). Table 3 presents the resulting dimensionless components for , and . The results demonstrated that the repeated image method model must include at least the first-order reflection term to satisfy the required computational accuracy for the electric field intensity. Table 3Dimensionless component.The number of reflections n 00.6134260.5443370.5332100.54146111.0089710.9999900.9998181.00000821.0019881.0000000.9999861.00001131.0006971.0000000.9999971.00000541.0005191.0000000.9999991.00000451.0000001.0000001.0000001.000000 Dimensionless component. Comparison of numerical solution and analytical solution To preliminarily verify the correctness of the analytical model, a four-layer numerical model with identical parameters was established based on the experimental conditions. The model employs the finite volume method (FVM) to solve Laplace’s equation on a Cartesian grid. The core principle of the FVM lies in discretizing the continuous computational domain into a series of non-overlapping control volumes that fully cover the domain. By integrating the conservation law over each control volume, a system of discrete algebraic equations is constructed to approximate the solution of the original equation. This method strictly ensures both local and global conservation, meaning that the net flux into any cell must equal the flux out of its adjacent cells. In other words, each discrete equation represents the rate of change of a physical quantity within a specific control volume being equal to the net flux entering through its boundaries plus any internal source terms42. Figure 10 shows the numerical model with grid discretization, where the base grid resolution is 2 cm × 2 cm × 2 cm. The grid near the electric current source was locally refined to a resolution of 2 cm × 2 cm × 1 cm. For comparison, two cases with different ice thicknesses were selected to calculate. The comparison between the analytical and numerical solutions is presented in Figs. 11, 12, 13 and 14, where the x-axis represents the index of measurement points along a single survey line, the y-axis indicates the case index, and the z-axis corresponds to the electric field strength. Numerical solution mesh partitioning. Comparison of the numerical solutions and the analytical solutions on Plane 1. Comparison of the numerical solutions and the analytical solutions on Plane 2. Comparison of the numerical solutions and the analytical solutions on Plane 3. Comparison of the numerical solutions and the analytical solutions on Plane 4. To more intuitively reflect the fitting accuracy between the numerical and analytical solutions, the relative deviation is defined as43: where denotes the analytical solution of the electric field intensity, denotes the numerical solution of the electric field intensity, denotes the index of measurement points along a survey line, and denotes the total number of measurement points on a single survey line. Table 4 presents the relative deviation values for 12 survey lines. Table 4The relative deviation table between the numerical solutions and the analytical solutions.PlaneSurvey lineThe relative deviationCase2Case4Plane 110.94%1.05%21.09%1.03%31.49%1.54%Plane 211.56%1.63%23.21%3.10%32.26%2.40%Plane 310.54%0.56%28.06%8.06%32.76%2.77%Plane 411.55%1.58%24.51%4.59%32.45%2.58% The relative deviation table between the numerical solutions and the analytical solutions. As presented in Table 4, the relative deviation between the analytical and numerical solutions across the 12 survey lines ranges from a minimum of 0.54% to a maximum of 8.06%. Among these, survey line 2 on Plane 3 exhibits a particularly large relative deviation. This discrepancy is attributed to the extremely small distance between this line and the sources, where the analytical solution indicates extremely rapid variations in field strength. The mesh resolution employed in the FVM is insufficient to fully capture such steep gradients, resulting in a numerical solution that overestimates the analytical reference at this specific location and leads to a higher deviation. Overall, the numerical model provides preliminary validation of the correctness of the analytical model. Comparison of analytical solution and measured value Comparative results between measured values and analytical solutions are presented in Figs. 15, 16, 17 and 18, where the x-axis represents the measurement point index along individual survey lines, the y-axis corresponds to the survey line index, and the z-axis denotes the magnitude of the electric field intensity. Comparison of the measured values and the analytical solutions in Case 1. Comparison of the measured values and the analytical solutions in Case 2. Comparison of the measured values and the analytical solutions in Case 3. Comparison of the measured values and the analytical solutions in Case 4. To more intuitively reflect the fitting accuracy between experimental values and analytical solutions, the relative error is defined as43: where denotes the experimentally measured values along a single survey line, represents the corresponding analytical model predictions at measurement points, indicates the measurement point index, and designates the total number of measurement points on a survey line. Table 5 summarizes the relative error values for all survey lines under the four cases: Table 5The relative error table between the measured values and the analytical solutions.PlaneSurvey lineThe relative errorCase1Case2Case3Case4Plane 119.87%9.75%8.87%8.84%211.23%11.28%7.53%7.67%36.03%5.76%5.24%5.38%Plane 214.41%3.39%6.20%5.25%22.58%2.85%3.13%3.20%32.15%1.98%1.68%1.61%Plane 311.34%1.47%1.63%2.25%23.46%3.51%3.54%3.87%31.64%1.48%1.37%1.17%Plane 415.22%5.19%5.38%5.69%28.60%8.51%8.62%9.42%34.08%3.86%4.38%5.14% The relative error table between the measured values and the analytical solutions. Discussion Overall, the relative error in Plane 1 is slightly higher than that in the other three planes. Plane 1 lies immediately adjacent to the sea ice–seawater interface, and the observed discrepancy is primarily attributed to distortion in the electromagnetic field distribution caused by boundary effects at the sea ice–seawater interface. According to the electromagnetic theory, the core principle of the repeated image method is to simplify the electric field calculation by replacing the irregularly distributed induced charges at the interface with equivalent image charges. Therefore, as the actual charge approaches the interface, its distribution deviates from the idealized image equivalence, which results in the increasing modeling error. Moreover, the sea ice layer in the experiment is a crystalline composite containing inhomogeneities such as air bubbles, which results in a spatially non-uniform distribution of electrical conductivity. This inherent heterogeneity diverges from the homogeneous medium assumption adopted in the analytical model, which further exacerbates the error in the electric field distribution near the interface. In Plane 3 and Plane 4, the relative errors observed along Survey Line 2, located along the x-axis, are generally higher than those recorded on Survey Line 1 and Survey Line 3. This is because Survey Line 2 is geometrically closest to the electric current source. Under these conditions, even minor deviations in the position, orientation, and excitation strength of the source will be amplified, which will result in noticeable variation in the field distribution and thus leading to large measurement errors. Furthermore, as can be seen from most survey lines, the relative error decreases slightly when the excitation current increases from 30 to 50 mA. The results aforementioned indicate that the random measurement noise can be suppressed to some extent by increasing the signal intensity, which improves the signal-to-noise ratio. Moreover, the comparative results validate the high reliability of the experimental parameters adopted in this paper, which can provide strong signal support for subsequent electric field measurements in low-temperature marine environments. In summary, under the four different cases involving varying ice thicknesses and current magnitudes, the relative errors across all survey lines were observed to range from a minimum of 1.17% to a maximum of 11.28%. The results aforementioned validate the effectiveness of the analytical solution model, which provides strong theoretical support for the rapid and accurate analysis of electric fields under sea-ice-covered maritime conditions. The electrical conductivity differences between the sea ice and seawater layers significantly modulate the reflection coefficients at the interfaces, which consequently governs the spatial attenuation and field distribution patterns. This demonstrates that incorporating the ice layer as a discrete environmental characteristic is essential for a reliable analysis of electric signatures in polar regions. Consequently, the proposed model enables the rapid and accurate calculation of electric fields in ice-covered maritime conditions. The model performs significantly faster than traditional numerical methods such as FEM or BEM. It offers robust theoretical support for the underwater electric stealth design of ships and the evaluation of anti-corrosion measures. Conclusion To investigate the distribution characteristics of electric field strength in ice-covered marine environments, this paper presented a four-layer medium model comprising air, sea ice, seawater, and seabed. Based on electromagnetic theory, the repeated image method was employed to derive an analytical solution for the electric field distribution within this stratified structure. The accuracy of this analytical model was verified through a comparative analysis with a finite volume method (FVM) numerical model, yielding maximum and minimum relative deviations of 8.06% and 0.54%, respectively. Furthermore, laboratory experiments provided empirical validation, demonstrating maximum and minimum relative errors of 11.28% and 1.17% when compared to the analytical solutions. Despite being characterized by computational efficiency, the current analytical model is limited to strictly parallel layers and controlled laboratory conditions, which may not fully capture the complexities of non-parallel interfaces or environments with significant 3D effects, such as irregular seabed topography, non-uniform sea ice thickness, and the heterogeneous distribution of ice conductivity in actual maritime environments. Additionally, while the laboratory experiments validated the model using scaled-down sea ice thicknesses (5–8 cm), the mathematical framework of the analytical model is theoretically applicable to a wide range of actual polar sea ice thicknesses (from several centimeters to several meters), provided the electrical conductivity can be approximated as a homogeneous effective value. In practical electrical prospecting applications, sophisticated filtering and rejection processing are required for multiple noise sources, including vessel corrosion electric fields, equipment leakage currents, and environmental noises from sea ice freezing-thawing, salinity gradients, marine organisms and seawater turbulence. Future research will focus on developing a multi-current source model for vessel underwater electric signatures in ice-covered seas based on the proposed analytical model, analyzing the impacts of layer thickness and conductivity variations on electric field distribution, and identifying equivalent source parameters with measured data. By accurately reconstructing electric signatures under specific polar conditions, the validated model will directly contribute to real-time vessel hull corrosion monitoring and anti-corrosion effectiveness evaluation. Moreover, it provides a critical foundation for non-contact sea mine risk control, helping to protect vessels against underwater detection systems triggered by electric field intensity measurements in ice-covered regions. Author contributions X.C.: Conceptualization, methodology, software, data curation, writing- original draft. Z.S.: Investigation, supervision, software, validation. K.Z.: Supervision, software, validation, writing-review and editing. M.L.: Visualization, writing-review and editing. R.J.: Software, validation. W.Y.: Software, validation. K.Y.: Visualization, writing-review and editing. Y.D.: Writing-review and editing. All authors reviewed the manuscript. Funding This work was supported by the National Natural Science Foundation of China (Grant No. 52202427). Data availability The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.