Source modification for efficiency enhancement of marine controlled-source electromagnetic method


 Controlled-source electromagnetics is a strongly efficient technique to explore deep-water marine hydrocarbon reservoirs. However, the shallow-water unsolved limitations of electromagnetic shooting methods still exist. In this regard, this work aims to alter the existing conventional electromagnetic source such that it can converge the down-going electromagnetic wave while simultaneously dispersing the up-going electromagnetic energy to minimise the airwave in shallow water. This work presents computed electric current distribution inside a modified transmitter, using a method of moments. Simulation and an experiment-based methodology are applied to this work. Finite element simulation of the response of the modified transmitter displayed the capability of the new transmitter in dispersing the airwave, by 15%. The experimental setup confirmed a better performance of the new transmitter, showing hydrocarbon delineation of up to 48%, compared to the existing conventional transmitter, with 25% oil delineation at the same depths in the same environment. Modification of the electromagnetic source to unbalance the up-down signals may have the potential to enhance the delineation magnitude of the target signal and, as a result, significantly improve oil detection capability.


Introduction
Low-frequency electromagnetic (EM) methods have been explored for the last few decades for offshore underwater investigations, onshore marine geological study and communication (Simkó & Mattsson 2004;Weitemeyer et al. 2006;Puzyrev et al. 2017;Hashmi et al. 2018). Cox et al. (1986) introduced an extremely low-frequency EM data collection method to maximise the penetration of the EM signal. They reduced background noises to as low as 1 pVm −1 , using a 1 Hz frequency. They reported that receivers at large offsets can detect a low-frequency EM field due to low attenuation. This work was a major breakthrough for EM methods in geophysical investigations (Constable 2010).
The seismic technique, while a powerful tool for the investigation of geophysical formations, struggles to distinguish hydrocarbon (HC) and water reservoirs (Low & Ghosh 2005). This is the reason why the certainty of seismic data in finding a marine HC reservoir is only 10-30%. Therefore, Ellingsrud et al. (2002) and Eidesmo et al. (2002) elevated marine controlled-source electromagnetic (MCSEM) technology to compliment the seismic technique for subaquatic HC exploration. Although uncertainty of the efficiency of the CSEM tool for risk reduction of detecting marine HC reservoirs still exists, the number of drilled wells in marine fields with an EM survey is significantly high (Hesthammer et al. 2010;Hesthammer et al. 2012;Alcocer et al. 2013;Stefatos et al. 2018).
Different alignments and methods for using electric dipoles as a controlled source have been used (Yahya et al. 2012;Helwig et al. 2013;Hashmi et al. 2018). Edwards & Chave (1986) used a horizontal electric dipole (HED) for resistivity investigations of sub-sea regions. The method was successful at imaging the resistivity of the sub-ocean to the order of tens of kilometres in depth. MacGregor et al. (2001) used the same method to study a resistivity map of the mid-ocean ridge.
The data produced by any EM tool, as well as CSEM, is dependent on electrical resistivity of geological strata (Constable 2010; Streich 2016; King et al. 2019). The electrically resistive layers are mainly either shales or HC reservoir. Shale is a consolidated dense sedimentary rock with a negligible fluid permeability. The seismic tool, as a density dependant tool, gives a clear image of shales after data processing of the formation, unlike liquid reservoirs (Alvarez et al. 2017;Shearer 2019). The resistivity of the HC reservoir, depending on the oil saturation, can be between 30 and 500 Ωm. HC reservoirs, however, are surrounded by low-resistive (1-2 Ωm) formations including rocks and sediment saturated by electrically conductive seawater. Figure 1 displays the CSEM wave components in marine geological layers using an electric controlled source. These components are classified into three major categories: direct wave, airwave and guided wave. Direct waves are emitted directly from the source to the receivers. The attenuation of the direct wave is exponential (Løseth & Amundsen 2007;Yahya et al. 2012). Therefore, the effect of this component at mid-offsets is negligible. The airwave is produced by the upward propagated EM signal inside the seawater (Løseth et al. 2010). The up-going wave travels at the interface of seawater and air, due to their large contrast of resistivity. The airwave is then transmitted back to the seawater and is detected by the array of receivers. The wave guided by the interface of the two media is attenuated only by a factor of r 3 (Kong et al. 2010). Therefore, the airwave affects even the large offsets. The guided wave, which is produced by the down-going wave, indicates the presence of a HC reservoir. The downgoing EM field is guided by the interface of the HC reservoir and its surroundings, and then reflects to the seabed and is recorded by the receivers (Yahya et al. 2012). The presence of a highly resistive reservoir affects the amplitude and phase of the recorded data (Kong et al. 2010;Rostami et al. 2016b). This can be interpreted as the HC delineation.
The antenna with a negligible cross-sectional radius in comparison to its length is considered a thin antenna. Method of moment (MoM) is a well-known technique for computational analysis of thin antennas. MoM was initially developed by Richmond and Newman (Richmond 1970;Richmond & Newman 1976). Afterwards, several researchers developed and implemented it for thin wire antenna analysis (Li et al. 2003;Shoory et al. 2003;Figueroa et al. 2004;Ozzaim 2018).
Since late 1970s, MoM has been applied to several geophysical studies concerning EM measurements and loggings. Wait & Hill (1979) used MoM to investigate the current distribution in a drill rod EM wave emitter located in a conducting rock medium. Implementing Pocklington's integral equation, DeGauque & Grudzinski (1987) studied lowfrequency antennas' current distributions of a conductive drill string surrounded by a conductive environment. EM propagation in a conductive medium was also studied by Xia & Chen (1993), by investigating EM signals receivability in a measurement-while-drilling system. Wei et al. (2010;Wei 2013) computed the current distribution in a vertically positioned thin antenna in a multi-layered system.
The HED transmitter is incapable of lessening noise due to the equal magnitude of the propagated up-going and down-going EM flow causing the airwave. On the basis of strong airwave noises in shallow-water regions, and the incapability of an HED transmitter to delineate shallow-water reservoirs, we modified the antenna. Thereby, we proposed a novel curved electric dipole (CED) as an electromagnetic source to disperse the up-going wave signal and converge the down-going wave. We applied MoM to compute electric current distribution in a low-frequency thin CED since the EM source is surrounded by conductive seawater. Furthermore, using a finite element solver, we addressed the capability of CED to reduce the scale of the airwave. In addition, we performed an experimental validation on the CED performance in a scaled model in a scaled tank using scaled transmitters with carefully scaled-down electrical input.

Method
The methodology of this work consists of the design of a new antenna, EM analysis of the designed CED using MoM, finite element-based simulation for CED performance in an oceanic EM survey and a scaled-down experiment to compare the performances of the conventional HED and the proposed CED sources. Data collected in experiments are analysed for comparison with the data from simulations to determine the correlations. Until a certain level of correlation is achieved then the simulations are deemed a qualified representation of the experimental data.

Development of the new transmitter
The idea of curved electric source to obtain more directivity was adopted from the directive transmitter concept that has been established for the wireless communication technology (Piesiewicz et al. 2008). The directive antenna is usually used in radar applications to obtain long-distance communication.
Here, for a fixed length of the transmitter, different curvature degrees were designed to acquire different directivities.
The horizontal electric source was curved with five gradual proportionate ratios of its height (h) and the radius (r) of the curved segment starting from horizontal to establish an arc with ratio (h = 0.5r). Equation (1) indicates the relation between length, height and radius of curvature. The geometries of the curved transmitter are displayed in figure 2a.
where L is the arc length, r is the curvature radius and h is the curvature height, introducing c = h/r as the curvature degree. Here, 'c' was taken from 0 (HED) to 0.5, at 0.1 intervals. Thereby, five curved electric transmitters are proposed. The Maxwell's equations can be expressed in terms of transverse electric (TE) and/or transverse magnetic (TM) fields. The magnetic field can be written as: where Π and Ψ represent the TM and TE modes, respectively. The current density of the source can be obtained by where J S z is the vertical component of the current density, and T and Υ are scalar functions that can satisfy the Poisson equation, which can be expressed as follows: The current density for a curve element, illustrated in figure 2a, is obtained by an integral equation as follows: where I is the electric current, R is the radius of the curve segment and is the angle of the curve element in polar coordinate that defines the curve segment, varying from 0 to 0 + (figure 2a). The TE and TM modes scalar functions in equation (2) can be obtained by partial differential equations as follows: where is the angular frequency of the source and and are the magnetic permeability and electric conductivity of the media, respectively. To obtain the produced magnetic 3 field via a curved transmitter, Fourier transform should be applied to equations (3-5). By substituting the inverse transforms in equations (6-8), we obtain where is the Green's function for the scalar Π, k is the wave number, the uppercase 0 on the Green's functions denotes the respective function with k = 0, J 0 is Bessel function and is the displacement from the infinitesimal curve segment to the observation point.
Equations (9) and (10) are the TE and TM mode expressions of a curved EM source in terms of the geometric parameters of the curved wire (R, ) depending on the curvature degree. Numerical solution to these equations will lead to obtain the magnetic field produced by a CED.

Method of moment for a curved thin transmitter
MoM, in electromagnetic, is a numerical solution to electric field (E-field) integral equations (EFIE) that can be stated in a linear operational form (Gibson 2014), where L is a linear integral form operating on an unknown field, f, to come up with a known vector field, E. The first step is discretising the function f on the domain of operator L, in a series of a linear form: where ɑ i s are unknown coefficients and f i s are basis functions. Substituting equation (12) in equation (11) we obtain The next step is to define weighting/testing functions ( 1 , 2 , … , N ) in the range of L that should be linearly independent. Applying inner product of i s into both sides of equation (13) where Z ij = ⟨w j ,L(f i )⟩, I i1 = i , and V i1 = ⟨w j , E⟩. Taking Z as the impedance matrix, I the matrix of unknown coefficients can be described solving the above equation as (Garg 2008;Nag & Datta 2015) [ To determine the unknown electric current on an arbitrary shaped wire, EFIE is used. The fact that the tangential component of the E field approaches zero on the surface of the transmitter obtains (Mei 1965): where r is the observation point's position vector on the surface of the wire, S. E i tan and E S tan are the tangential components of the incident and scattered E field, respectively. The E field can be expressed as a function of magnetic vector potential A(r), where A(r) is a function of the current distribution I(r) in a wire filament with radius a(r) (Barrera- Figueroa et al. 2005). Thus, we obtain Since I(r) is not a function of azimuth and is only dependent to the arc length, for a thin wire filament, this can be expressed as I(s ′ ). Thus, this can describe the position vector as a function of the arclength, s.
wheren is the unique vector normal to the wire's alignment. Introducingŝ, the unique vector of the wire's axis andŝ ′ , the unique vector of the electric current filament and considering the change in variables of equations (18), equation (17) obtains a linear integration to express the E field on the surface of the wire as follows: whereŝ is the unique vector align wire's axis andŝ ′ is the unique vector parallel to the electric current filament. Since equation (19) is a linear EFIE, we can use MoM to transform it to a matrix form. As a good approximation, I(s ′ ) can be expressed as a series of linear basis functions as where I n s are unknown coefficients representing the current on each segment and the n (r)s are the known linear basis functions, defined over each arc segments. Substituting equation (20) into equation (19) yields Weighting function. To obtain the best solution for EFIE, selection of an appropriate testing function is inevitable. The testing function is expressed as the weighting function for the integral equations. To obtain independent expressions for each unknown, inner product of the weighting function is applied to the EFIE (equation 21).
Considering elements of electric potential matrix, v m = ∫ s w m Eds, we have where, [Z mn ] is the impedance matrix and [I n ] is the matrix of unknown coefficients expressing the electric current of the thin wire. Elements of the impedance matrix can be written as (24) Following a point matching technique, Dirac's delta function is applied to equation (24) to express the weighting function, w m . A piecewise linear function is chosen as the basis function.
Applying the basis function in equation (25) into (24), the impedance becomes Source modelling. Unlike the conventional dipole antenna, the dipole transmitters used in MCSEM are end-feed dipoles. Therefore, we used a delta gap function source for the two ends of the CED, i.e. the first and the last segments. The impressed E field at the transmitter's terminals is expressed as where Δ s is the width of each terminal and V 0 is the initial input electric potential, connected to the first and N th (last) segments, since the value of the initial voltage for the rest of the segments is zero. Thus, the matrix of electric potential of the source becomes If the impedance of segments 1 and N are Z and Z' , respectively, potential difference of ends of these segments can be obtained Substituting equation (29) into (23) we have (30) It can be seen that connecting an impedance to the i th element is equivalent to adding the same value to the matrix element Z ii . Having the source and impedance matrixes, we can obtain unknown elements of the matrix I, as explained in equation (15).

Simulation
The low-frequency (LF) module in the Computer Simulation Technology Electromagnetic Studio (CST EM Studio) (CST Computer Simulation Technology version 2012) was used to simulate the response and propagation of the E field in an inhomogeneous or layered environment. For CSEM modelling, the LF module of CST EM Studio was used. Its background is based on a finite element method to solve a partial differential form of Maxwell's equations for the LF approach of electromagnetic wave propagation. The simulation takes into account the resistivity properties of each of the inhomogeneous or layered environments by applying Maxwell's equations and the theory of resistivity. Three main EM parameters describe the EM properties of the model: electric permittivity , magnetic permeability and electric conductivity . With the angular frequency, , wave number is defined to describe the propagation: where, c is the wave velocity, = 2 f is the angular frequency and is the skin depth. The skin depth is the distance that an EM wave can travel with less than 63% of attenuation. In equation (31) the first term is an indication of displacement current, from Maxwell's equations, and the second term describes the conduction current. Therefore, equation (31) depicts that the displacement current is negligible in LF approach, where ≫ . The model is built to simulate the shallow-water region for a EM geophysical survey. In shallow waters, the airwave is the most dominant recorded signal. A region of 25 × 25 km and 550 m of seawater depth is modelled. The electric source is placed 30 m above the seabed, at the centre of the horizontal offset. Figure 3 illustrates the geometric parameters of the simulation model.
The generated EM response is dependent on the magnetic permeability, electrical conductivity and electric permittivity of different layers, such as the HC reservoir, sediment, seawater and air. The model consists of five different layers. Layer 1 is air, with a negligible electrical conductivity of 1.0 × 10 −10 Sm −1 , layer 2 is seawater with varying electrical conductivity due to its salinity, layers 3 and 4 are the over-burden and under-burden that were assumed as conductive brine saturated porous rocks and layer 5 is the resistive HC anomaly. The three main EM properties of the simulated CSEM model are indicated in Table 1.

Scaled-down experiment
A huge inbuilt scaled-tank setup was used to simulate MC-SEM environment, as displayed in figure 4a. The scaled tank with a cross section of 2 × 9 m and height of 2.2 m is filled by low-resistive saltwater and a layer of saltwater saturated sand. The sand layer, representing the reservoir's over-burden and under-burden, has a thickness of 0.6 m and the thickness of saltwater is 0.5 m. Saltwater is made from the dissolution of plane water with cooking salt at 35 g l −1 of water. The oil layer was prepared with the Perspex cubic containers of 6 mm thickness and dimensions of 1.80 × 1.20 × 0.12 m; filling them with bags of cooking oil. The sandbags and Perspex containers in the scaled tank are displayed in figure 4c. An aluminium rod 1.41 m long was used as the electric source. The CED transmitter was established by bending the aluminium rod with the same properties as the HED transmitter. Here, we tested a CED of c = 0.3 curvature degrees. Three fluxgate magnetic field sensors Mag-0xMSS100 were placed at three different positions in the scaled tank, on the surface of the sandbags. A 20 V P-P square wave was applied to the electric source by an alternating current signal generator. The electric source was deployed slightly above the sand, at a maximum height of 15 cm, at a low speed of 0.0067 m s −1 . The resistivity measurements of different layers such as oil and saltwater were carried out by a MC-MR-III resistivity meter. The magnitude of the magnetic field of propagated EM energy against the transmitter-receiver offset was recorded by a NI PXI-1042 deca-port data acquisition system. The resistivity values and experimental parametric values are shown in Table 2. Further details on the experiment parameters, i.e. real and laboratory-scale dimensions, are given in Table 3.

Electric current distribution in thin CED
Electric current distribution was computed using MoM for EFIE. Figure 5 shows the electric current distribution in CED transmitters with different curvature degrees, from c = 0.1 to c = 0.5 (see figure 2b), against the length of the arc. The CEDs are designed for MCSEM application. Therefore, the arc length that is chosen for marine electromagnetic survey is 270 m.
The two terminals of the curved wires represent the maximum current since the CEDs are end-feed transmitters. The current distribution slightly decreases towards the centre of the CEDs. The small difference of current distribution versus curvature degree of CEDs indicates that the current distribution is not disturbed by the shape of transmitter. This is due to the LF approach for a wire transmitter (Cagnoli, Gammaitoni et al. 1999). In the other words, in the LF domain there is less dependency of the distributed electric current inside the wire on the shape of the wire, although the wire's geometry might affect the generated field by the current carrier. This concludes that performance of the LF electric dipole, used in marine survey applications, will not be significantly affected by bending the transmitter.

Modelling of the CED for shallow-water survey
We used the LF module of the CST EM Studio finite element solver to investigate the performance of the modified antennas in (i) airwave dispersion and (ii) marine hydrocarbon exploration.

Dispersion of airwaves in an MCSEM inhomogeneous model.
In the model displayed in figure 3 the amount of airwave dispersion was computed using HED and five CEDs at c = 0.1, 0.2, 0.3, 0.4 and 0.5°of curvature. An extremely LF (0.125 Hz) alternating current with an amplitude of 1250 A was applied to the electric source. To observe the generated airwave using HED and each of the CED sources, the E-field response was collected from the seawater surface. The plot of the E field against offset is shown in figure 6a. To have a clear illustration of the performance of each transmitter, figure 6b displays the magnification of the E field at the offset range of 5-15 km. The up-going wave produced from HED and three different CEDs of 0.1, 0.2, 0.3, 0.4 and 0.5 curvature degrees was compared. According to figure 6, the sizes of airwaves that were produced by up-going wave guided by seawater surface decreased with the increase in curvature degree of the electric source. This amount is the minimum for the CED of c = 0.5 that can be translated as airwave dispersion. Thus, the airwave dispersion of CED of 0.5 curvature was at the maximum, showing 15% more than that of the HED. Table 4 indicates the E-field response at different offsets.

Exploration of hydrocarbon reservoir in shallow-water region.
In this section, the modified electric sources are tested to explore HC in the presence of strong airwave signals in a shallow-water region. The HC resistive layer that was 150 m thick was modelled at 1.0 km deep below the seabed. The magnitude of the E field versus the offset for the two models with and without the HC layer are plotted together. The difference in the E-field magnitude between these two models, illustrated in figure 3a and b, indicates the presence of the resistive HC reservoir in the model. The percentage of the HC delineation is calculated based on the difference between Efield responses of the models with and without the presence of an oil reservoir.
where E withHC and E noHC are the E-field signals of the models with and without HC layers, at a certain offset. Figure 7 illustrates the magnitude of the E field against offset, generated by HED and each CED of 0.1 to 0.5 curvature degrees.
Results from Table 5 show that the CED source of 0.3 curvature degrees is the most efficient electric source for HC exploration in a shallow-water region. It can also be concluded that for HC exploration in presence of strong airwave, the performance of 0.3 CED, with 35% delineation, is improved by 10% compared to that of HED with 32% delineation. It is worth mentioning that the simulation was performed without noise filtration techniques to evaluate only the impact of curvature in MCSEM survey. Using noise cleaning methods may enhance the efficiency of the CED even more.

Experimental simulation: the scaled tank
The scaled-tank experimental setup aims to compare the performances of the modified CED source with the conventional HED. The experimental setup is displayed in figure 4. To calculate the frequency of laboratory-scale experiment, scale factor is taken into account Rauf et al. 2019). At low frequencies, equation (31) can be written as The skin depth and the wavelength can thus be obtained as To define the scaling for the experiment, skin depth is taken into consideration. Thus, the scale factor, which is the ratio of the dimensions of the laboratory scale to the real-scale dimensions, can be obtained by the ration of laboratory-to real-scale skin depth, as given by equation (36). Equation (37) indicates the laboratory-scale frequency formula, obtained from equations (34) and (36).
where n (in our case = 0.0004) indicates the scale factor, d real and d lab are the dimensions of real and laboratory scales, respectively, f real is the applied frequency used in real scale and f lab is the frequency of laboratory scale. Taking real-scale frequency as 0.125 Hz, the frequency used in the laboratory will be 780 kHz. The collection of raw data from the tank was done in three distinct stages. The first stage was to collect the con-trol data set. This was done without any current flow fed to the transmitter. The transmitter and receivers were laid in situ to collect ambient data for a predefined period. This data set was then modelled to represent the telluric noise, noise from the environment and unexplained sources. This was the first noise to be eliminated from the experimental data.
The second data set was collected with the transmitter and receivers in situ with the current fed to the transmitter. This set included the telluric noise and the other noises emanating from the reflection of boundary walls and other reflective materials within the bounds of the setup. This defined the second set of data to be eliminated from the raw data after the experiment was run.
The third and most important raw data set was collected while the transmitter was powered on and towed across the simulation setup to record responses including reflections from the oil bags. This data included the environmental noise collected in stage one, the telluric noise collected in stage two 11 Figure 9. B-field MVO, using CED in three Cartesian directions: (a) x, (b) y and (c) z, recorded by sensor 2. and the data from the oil bags. This sequence enabled the quantification and characterisation of different signal components within the final data set. Quantification and characterisation of signal components, in turn, allowed the identification and elimination of such components using wavelets and/or independent component analysis-based methods.
The response of the magnetic field (B-field) components was recorded by three magnetic receivers located inside the scaled tank, as illustrated in figure 4a. The distance of the first sensor from the oil container was at the maximum, the second receiver was placed at the centre of the scaled tank and the third receiver was placed exactly above the oil container. The large amount of direct wave signal masks the wanted signal, i.e. the guided wave. Therefore, only the recorded signals by the other two magnetic sensors are considered.
First, a horizontal aluminium rod was used as the conventional HED source. The movement of the electric source through the scaled tank was at a constant speed of 1 m min −1 . The magnitude versus offset (MVO) of the B-field response at three Cartesian directions was recorded by the magnetic receivers, as displayed in figure 8. The offset origin is the scaled tank's right edge.
The detected B-field response in figure 8i indicates that sensor 2 was more capable of delineating the guided wave in comparison with sensor 1. This is due to the further distance of sensor 1 from the source of the guided wave, i.e. the oil container. On the other hand, sensor 3 was located at the top of the oil bags. Therefore, the response of the guided wave from the oil-sand interface was masked by the highly intense direct waves. Thus, sensor 2, which is located at the centre of the scaled tank, was in the best position of delineating the buried oil layer. The logarithmic plot of B-field response represents a more explicit illustration of the MVO result. Therefore, the logarithmic view of the B-field against offset is given in figure 8ii.
Next, is to investigate delineation of HC using a CED of 0.3 curvature degrees. Figure 9 displays the received B-field signal by sensor 2.
Considering the origin of the offset at the right edge of the scaled-thank, shown in the schematic diagram of figure 4, the peaks in MVO graphs indicate the direct wave signal. This signal, at the middle of the offset, is recorded by the magnetic sensor placed at the centre of the scaled tank. The anomalies in the right side of each MVO graph are an indication of the guided wave. The guided wave was produced due to the presence of the resistive oil container. Thereby, the offsets of anomalies determine the position of the resistive HC layer.
The percentages of delineation of HC using conventional HED and new CED sources are mentioned in their respective MVO graphs. The stronger HC delineation of the CED is due to its capability of dispersing the up-going wave, leading a reduction in the airwave effect. At the same time, CED is able to converge the down-going EM energy, which provides the guided wave as the wanted signal. 12 Figure 10. B-field vs offset, using CED and HED in three Cartesian directions: (a) x, (b) y and (c) z, recorded by sensor 2.
Comparing figure 8ii and figure 9 indicates that the modified electric sourced is more efficient than the conventional straight transmitter for detecting the resistive anomaly. Figure 10 represents the comparative B-field MVO graphs obtained using HED and CED of 0.3 curvature degrees. Figure 10 shows that the anomaly due to presence of HC generated by the CED source is more explicit compared to the conventional HED source. The fact that a conventional dipole antenna produces an omni-directional pattern, indicates that it propagates an equal amount of energy in all directions. This illustrates the fundamental shortage of the conventional HED used in CSEM survey in that it produces an equal up-going energy as the down-going waves. The airwave that is produced by the up-going waves is a significant noise versus the guided wave response produced by downgoing waves. Thus, airwave and guided wave interaction is a destructive interference that hides the effect of the guided wave for hydrocarbon delineation. Therefore, using a CED, which converges the down-going wave and disperses the upgoing energy, performs significantly more efficiently than the straight dipole for delineating the buried oil layer. Table 6 shows the resistive oil delineation using the conventional transmitter and the modified CED source. Tabular results show a stronger performance of the curved electric source in comparison with the HED transmitter. The delineation percentage using the CED transmitter is more than double that obtained by the HED source. The performance of the new designed CED transmitter can be a breakthrough for marine electromagnetic methods for hydrocarbon exploration. However, the deployment challenges of such a transmitter in the oceanic environment should be taken into account. One recommendation is to use the CED as a static electric source. After collecting data from a predefined zone, the source can be relocated to the next predefined data collection zone. This prevents the difficulty of a continuous towing of a curved structure/huge cable. However, to design a setup that can carry a long CED inside the seawater can be a way forward to this work.

Conclusion
CED was introduced as an electromagnetic source for marine hydrocarbon exploration surveillance. We computed the electric current distribution of the modified transmitter with different curvature degrees to validate its electromagnetic behavioural performance. Next, an inhomogeneous layered SBL model was simulated in CST EM Studio to evaluate 13 airwave dispersion of different curvature degrees of the CED in a shallow-water environment. The CED of c = 0.5 represented the highest airwave dispersion. In addition, exploration of HC was tested for the conventional straight transmitter and the modified electric source. The CED of c = 0.3 showed the maximum delineation of 35% for a marine HC reservoir in a shallow-water region, whereas the HC delineation of HED was 32%. The experimental validation of CED indicated the significance of its performance where the 0.3 CED delineated HC by 48%, which is more than double the HC delineation of the HED.