Airborne Electromagnetic Bathymetry

Traditional methods for measuring the water depth rely on sonar soundings. However airborne techniques offer the advantages of increased survey speed and operation over dangerous waters that may be affected by very strong tides and the presence of shoals and reefs that limit the operation of surface vessels. The airborne lidar method has been used very successfully for mapping coastal waters and relies on the difference in the time between a laser pulse (infra red wavelength) reflected from the sea surface and a separate laser pulse (blue-green wavelength) reflected from the sea floor. The depth of investigation, typically 50 to 70 m in ideal conditions, depends strongly on water clarity (turbidity), as well as other factors (e.g. sea state, surf zone, sea bottom reflection) and weather conditions. These lidar systems provide dense depth soundings, typically a grid with a laser spot spacing of 4 to 5 m, and they meet the accuracy standards of the International Hydrographic Organisation (IHO).


Introduction
Traditional methods for measuring the water depth rely on sonar soundings.However airborne techniques offer the advantages of increased survey speed and operation over dangerous waters that may be affected by very strong tides and the presence of shoals and reefs that limit the operation of surface vessels.The airborne lidar method has been used very successfully for mapping coastal waters and relies on the difference in the time between a laser pulse (infra red wavelength) reflected from the sea surface and a separate laser pulse (blue-green wavelength) reflected from the sea floor.The depth of investigation, typically 50 to 70 m in ideal conditions, depends strongly on water clarity (turbidity), as well as other factors (e.g.sea state, surf zone, sea bottom reflection) and weather conditions.These lidar systems provide dense depth soundings, typically a grid with a laser spot spacing of 4 to 5 m, and they meet the accuracy standards of the International Hydrographic Organisation (IHO).
The airborne electromagnetic (AEM) bathymetry method is based on the AEM technique (Spies et al., 1998;Palacky & West, 1991) developed for geological exploration of electrically conducting targets, initially applied to mineral exploration.Since then, the AEM technique has also been applied to environmental studies, e.g.mapping hydrogeological features in alluvial aquifers (Dickinson et al., 2010); and salinity distribution (Fitterman and Deszcz-Pan, 1998;Hatch et al., 2010;Kirkegaard et al., 2011).The AEM method uses an airborne transmitter loop with a known magnetic moment (i.e. a magnetic dipole source) that generates a primary magnetic field to induce electrical currents in the ground.These currents establish a secondary magnetic field (the EM response) which is detected by the airborne receiver loop, shown schematically in Figure 1.The mathematical theory of this EM induction process is thoroughly reviewed by Ward & Hohman (1987), Wait (1982), Weaver (1994), West & Macnae (1991) and Grant & West (1965).The EM induction process causes the electromagnetic fields to diffuse slowly into the conductive medium.Nabighian (1979) showed that the transient time-domain electromagnetic (TEM) response (caused by a rapid turn-off of the transmitter current) "observed over a conducting half-space 1 or a layered earth can be represented by a simple current filament of the same shape as the transmitter loop moving downward and outwards (in the isotropic conductive medium) with a decreasing velocity and diminishing amplitude, resembling a system of "smoke rings" blown by the transmitter loop" (Nabighian, 1979).This concept was later extended to a frequency domain EM response (Reid & Macnae, 1998) where the transmitter loop is powered by a continuous sinusoidal current at a pre-determined frequency.Thus the footprint associated with the EM induction process (to be discussed in section 3.4) is much larger than the lidar footprint.The physics of the diffusion of the EM fields in the conductive medium is such that for airborne (or ground) EM systems the horizontal resolution will always be larger than the horizontal resolution of lidar-based bathymetry systems because of the relatively small footprint of the spot laser beam.However, a distinct advantage of the AEM technique is that the EM response depends on the bulk conductivity of the medium (as well as other parameters), which in the case of seawater, is not affected by the turbidity or the presence of bubbles.Thus the interpretation of AEM data acquired over seawater is expected to provide water depths, in relatively shallow coastal waters irrespective of turbidity levels (within reasonable limits), and in the surf zone.
The simplest model for interpreting AEM data for bathymetry studies assumes a onedimensional (1D) layered-earth structure which in its crudest form consists of two upper layers, a seawater layer overlying an unconsolidated sediment layer, with each layer defined by its electrical conductivity and thickness; these layers in turn overlie a relatively resistive 2 basement (bedrock).This two-layer model allows for a sediment layer (which may become vanishingly thin over exposed bedrock on the seafloor) and assumes no stratification in the conductivity of the seawater and sediment layers.Distinct conductive layers of seawater and/or sediment can be readily included by introducing more layers if warranted.The parameters used for inversion of the AEM data are the two thicknesses of the upper two layers, the three electrical conductivities associated with the two layers and the basement, and the elevations (and separations) of the transmitter and receiver loops above the sea surface.The thickness of the upper layer, as determined from the inversion of the AEM data, is the water depth which is the objective of the AEM bathymetry method.Where some parameters are measured directly, for example seawater conductivity and altimetry, or assumed, for example, sediment conductivity and basement resistivity, then these parameters may be held fixed or tightly constrained in the inversion process.
Given that the sediment layer and its associated conductivity are part of the layered-earth model, the thickness and conductivity of the sediment layer may also be determined in shallow waters from the AEM data.Combined with interpreted water depths (i.e.AEM bathymetry), the AEM method can therefore be used to (i) estimate the bedrock topography by combining the water depth and sediment thickness (Vrbancich, 2009;Vrbancich & Fullagar, 2007a) and (ii) map the seafloor resistivity (Won & Smits, 1986a).Combined with empirical relationships such as Archie's Equation (Archie, 1942) and assumed cementation factors for unconsolidated marine sediments (Glover, 2009), the derived seafloor resistivities can be used to estimate important seafloor properties such as density, porosity, sound speed and from these properties, acoustic reflectivity (Won & Smits, 1986a).The interpretation of AEM data to determine sediment thickness and sediment conductivity requires a well calibrated AEM system.The EM response is far more sensitive to the overlying conductive seawater layer than it is to the less conductive sediment layer and it is therefore more difficult to accurately determine the depth of the sediment-basement interface than the depth of the seawatersediment interface.The same AEM dataset may provide bathymetric accuracy to within 1 metre and yet provide poor agreement with bedrock depths estimated from marine seismic data.The use of the AEM method for simultaneously mapping water depth and sediment thickness and conductivity is currently under investigation and for this purpose, it is important to have independent ground truth data consisting of (i) marine seismic survey data (matching the AEM survey line locations if possible) to provide an estimate of the depth to bedrock, and (ii) sediment resistivity data obtained, ideally, from bore hole sediment samples, or surficial sediment samples (extending to 3-5 m depth) obtained from vibrocore samples.A marine seismic survey and a resistivity study of vibrocore samples of shallow marine sediments has been undertaken in Jervis Bay (Vrbancich et al., 2011a) and Broken Bay (Vrbancich et al., 2011b), located approximately 150 km south and 40 km north of Sydney Harbour (~ 33.8º S, 151.3º E) respectively, to support the interpretation of AEM survey data to estimate depth of bedrock and sediment resistivity (Vrbancich, 2010).

Chapter outline
Section 2 discusses the initial development of AEM for bathymetric studies, and the associated application of AEM for sub-ice bathymetry and sea-ice thickness measurements.Section 3 describes helicopter and fixed-wing AEM systems, and some of the transmittercurrent waveforms as well as discussing the AEM footprint that is relevant for understanding the lateral resolution.Section 4 provides some examples of the results of AEM bathymetry studies from helicopter 3 time-domain and frequency-domain surveys in Australian waters, comparing the interpreted water depths with known values.Section 5 (Conclusion) summarises the important findings and briefly discusses future directions.

Development of the AEM bathymetry method -initial studies
Initial studies in the use of AEM for bathymetric mapping took place in the 1980s.Morrison and Becker (1982) were the first to consider the use of AEM systems for mapping water depths in a feasibility report commissioned by the US Office of Naval Research (ONR) to support rapid airborne bathymetric mapping in shallow coastal waters.Morrison and Becker (1982) investigated both time domain and frequency domain systems existing at the time and concluded that frequency-domain systems would be limited to a depth of approximately 20 m and that the INPUT4 fixed-wing time-domain system, with a lower operating base frequency, was suitable for measurement depths of 50 to 60 m.This feasibility study was followed by single flight line field trials in Nova Scotia and New Brunswick (Canada) using the INPUT system over waters that were predominantly less than about 20 m in depth, resolving water depths to about 40 m with approximately 2 m accuracy as determined by comparison with soundings on coastal charts (Becker et al., 1984;Becker et al., 1986;Zollinger, 1985;Zollinger et al., 1987).
During this period of study, Won and Smits (1985, 1986b, 1987a), and Son (1985), also investigated the use of helicopter frequency-domain AEM for bathymetric applications.Based on field trials conducted in the Cape Cod Bay area (Massachusetts, USA) using a DIGHEM III AEM system (Fraser, 1978), Won and Smits (1985, 1986b, 1987a) showed that excellent agreement in water depths was obtained with acoustic profiles down to depths of about 13 to 20 m corresponding to one and one and a half skin depths at 385 Hz (the lowest of the 2 frequencies used in the DIGHEM III survey).The same dataset was also used to derive continuous profiles of seawater and seabed conductivity (Won & Smits, 1986a, 1987b).Bergeron and co-workers (1989) examined the same Cape Cod Bay dataset using a different method to invert the AEM data, to obtain good agreement with measured altimetry, seawater conductivity and known water depths.Bryan et al. (2003) used field data obtained over marsh and estuarine waters in Barataria Basin (Louisiana, USA) with a frequency domain AEM system using a primary waveform digitally constructed from six harmonic frequencies (Mozley et al., 1991a,b).The results of their inversions showed good agreement with measured water conductivities and depths, identifying horizontal water layers with the less saline, less conductive, fresher water layer overlying a more saline, more conductive water layer.The same instrumentation was used by Mozley et al. (1991a,b) in field trials at Kings Bay (Georgia, USA) to successfully map seawater depths and conductivity, and variations in seafloor sediment conductivity, and by Pelletier and Holladay (1994) to map bathymetry, sediment and water properties in a complex coastal environment located at Cape Lookout, North Carolina, USA.

Airborne electromagnetic bathymetry and sea ice thickness measurements
The EM response is sensitive to sub-metre variations in the altitude of the transmitter and receiver loops above the conductive seawater layer.This sensitivity has implications for measuring sea ice thickness.If the AEM data is also inverted for altitude, then the thickness of any sea ice covering the seawater can be determined by subtracting the elevation above the surface of the relatively resistive sea ice (measured with laser altimetry) from the altitude above the seawater (Becker & Morrison, 1983;Kovacs & Holladay, 1990;Reid et al., 2003a;Haas et al., 2009).This method relies on the conductivity contrast between sea ice and seawater.As sea ice ages, brine is expelled decreasing its conductivity.The seawater conductivity is typically 2.5 -2.6 S/m, approximately two orders of magnitude larger than sea ice conductivity and at the relatively low frequencies used in frequency-domain AEM systems (typically 50 Hz to 100 kHz), the primary and induced secondary magnetic fields effectively "see through" the sea ice.The AEM system can therefore be used to measure sea ice thickness and also sub-ice bathymetry (Kovacs & Valleau, 1987;Pfaffling & Reid, 2009) in shallow waters.This same EM induction technique can also be applied using ship-borne sensors (Reid et al., 2003a;Reid et al., 2003b;Haas et al., 1997).Unlike the AEM bathymetry method which usually assumes a layered earth model for 1D inversion, determining the structure of three-dimensional sea ice keels (pressure ridges below the sea surface) with thicknesses that may reach 6 to 10 m may require the use of 2D and 3D EM modelling and inversion procedures (Reid et al., 2003a;Liu & Becker, 1990;Liu et al., 1991;Soininen et al., 1998) in order to minimise underestimating the thickness of the ice keels that results from smoothing of the EM response over the AEM system footprint (Kovacs et al., 1995;Liu & Becker, 1990) and interpretation using 1D inversion models.

AEM systems: Time-domain and frequency-domain
AEM systems operate in the frequency domain (frequency electromagnetic, FEM) or in the time domain (transient electromagnetic, TEM)5 .Helicopter AEM systems typically operate in either the frequency domain or time domain, whilst fixed-wing AEM systems typically operate in the time domain.Fountain (1998) has reviewed the first 50 years of development of AEM systems from a historical perspective.A receiver loop is used to detect the secondary magnetic field induced in the ground.The recorded response is a voltage proportional to the time derivative of a component of the secondary field, typically the vertical component (dB z /dt).In the following text, the term "bird" refers to the AEM system towed as a sling load beneath the helicopter in FEM and TEM systems, or to the receiver unit towed beneath a fixed-wing TEM system.All three types of AEM systems have been used for bathymetric investigations in Australian coastal waters.

Frequency domain helicopter AEM
The FEM system consists of several transmitter-receiver coil pairs held in a fixed rigid geometry.The transmitter-receiver coil pair separation is about 8 m or less, and the coils may be (i) placed horizontally (horizontal co-planar (HCP) configuration, vertical magnetic moment associated with the transmitter coil), (ii) or the coil pair may be rotated by 90°, so that the coils lie in the vertical plane (vertical co-planar (VCP) configuration, horizontal magnetic moment, orthogonal to the flight direction, associated with the transmitter coil), (iii) or the coil axes lie on the same horizontal line (vertical co-axial (VCX) configuration, horizontal magnetic moment in-line with flight direction, associated with the transmitter coil).Thus with respect to the flight direction, the HCP, VCP and VCX coil configurations have the transmitter dipole moment aligned vertically, transversely and longitudinally respectively with regards to the flight direction.The AEM data consists of the in-phase (R, real response) and quadrature (Q, imaginary response, relative to the primary field from the transmitter) signals detected by the receiver coils.The phase (φ) of the response is given by φ = arctan(R/Q).
The first AEM bathymetry survey in Australia took place in 1998 over Sydney Harbour (New South Wales, Australia) using a DIGHEM V system (Figure 2a), originally developed by Fraser (1978) for resistivity mapping of metallic mineral deposits in the 1970s using earlier DIGHEM versions.The DIGHEM V system is an analogue instrument6 consisting of 5 coil pairs (3 HCP and 2 VCX, Table 1).The lowest frequency (f) was tuned to 328 Hz to  (Vrbancich, et al., 2000a,b).The smaller bird between the helicopter and the DIGHEM bird is a magnetometer bird; (b): The HoistEM time-domain helicopter AEM bird, located over the Sow and Pigs reef during a survey of Sydney Harbour (Vrbancich & Fullagar, 2004, 2007b).The transmitter loop (~ 22 m diameter) is attached to the extremities of the poles and the multi-turn receiver loop is located on the same plane, at the centre of the system.maximise the depth of penetration through seawater, i.e., to increase the skin depth () where  (m) = 500/(f) 1/2 = 250/(f) 1/2 for a typical seawater conductivity () of 4 S/m.Following this survey, Shoalwater Bay (Queensland, Australia; Vrbancich, 2004) and Sydney Harbour have been surveyed using an analogue DIGHEM_Res(istivity) instrument with 5 HCP coil pairs operating within the range of 387 Hz to 103 kHz (Table 1 One advantage of FEM helicopter AEM systems (compared to fixed-wing TEM systems) is that the transmitter and receiver coils are contained in a rigid structure so that the coils are held in a fixed position relative to each other.However pendulum motion of the towed bird generates a geometric and inductive effect in the measured EM response and contributes to altimeter error (Davis et al., 2006).If the bird swing period can be determined from survey data, a filter can be desig ned to remove bird swing effects, to first order, caused by pendulum motion (Davis et al., 2009).Predictions of bird swing from GPS receivers mounted on the bird and the helicopter can be used to predict the bird maneuver (Davis et al., 2009;Kratzer & Vrbancich, 2007).Calibration errors in FEM helicopter data can be identified by transforming the data to several different response-parameter domains and used to minimise the effect of altimeter and bird maneuver errors (Ley-Cooper et al., 2006).Importantly, this procedure can be applied to historic data, i.e., previous/dated FEM helicopter AEM data can be re-analysed with this procedure.

Fixed-wing time-domain AEM
A TEM system does not use a continuous sinusoidal current to power the transmitter loop (as in FEM systems).Instead, typically, the transmitter current is increased to a maximum value (during the "on-time"), and then reduced to "zero" current and measurements are made during the "off-time" when the transmitter is not powered 7 .Two examples of TEM waveforms are shown schematically in Figure 3.The secondary magnetic fields related to the decay of the induced currents in the ground are detected whilst there is no primary field.The process is repeated using the same current pulse but with the opposite polarity and the same off-time interval (Figure 3), and this response is subtracted from the first response to improve the signal to noise ratio.The period of the waveform (consisting of the the two ontime periods (Δt 1 ) of opposite polarity and their corresponding off-time periods, Δt 2 ) represents the base frequency(1/Δt 3 ), Figure 3; 25 Hz is used in Australia and 30 Hz is used in America to minimise 50 Hz and 60 Hz power line transmission interference respectively.The results of the subtraction between the measurements made during the first and second halves of the waveform are stacked (averaged) over many cycles to reduce noise.remains.Morrison et al (1969) also noted that given that the skin depth is inversely proportional to the square root of the frequency, then the early part of the transient decay is governed by rapid decay of high frequency energy which has only penetrated down to relatively shallow depths whilst at later times, the response is dominated by lower frequency energy which has penetrated to greater depths.Resistive media are associated with a rapid decay, conductive media are associated with a longer decay.For bathymetric applications, early time (and higher frequency) AEM data are required for shallow seawater depths and late time (and low frequency) AEM data are required for deeper seawater.In principle, the lower the base frequency for TEM systems and the lower the frequency for FEM systems, then the greater the depth of investigation in seawater.

Variable transmitter-receiver geometry
Fixed-wing TEM systems have the transmitter coil spanning the wingtips and front and rear extremities of the aircraft, as shown schematically in Figure 1, with the receiver coil contained in a "bird" that is released from the rear of the aircraft.The transmitter and receiver operate at different heights above ground level.The aircraft survey altitude is typically about 120 m and the bird has an assumed fixed horizontal offset within the range of 90 to 120 m and an assumed fixed vertical offset within the range of 40 to 60 m, relative to the centre of the transmitter.Unlike FEM helicopter systems, the relative geometry between the receiver and transmitter is variable leading to interpretation errors arising from unrecorded variations of bird attitude, offset and altitude, thereby limiting the potential of the AEM bathymetry method.Vrbancich & Smith (2005) estimated bird position errors of several metres at survey altitude using GEOTEM data, based on the prediction of the vector components of the primary field measured by the receiver at high altitude, i.e. assuming a free-space approximation (Smith, 2001a) and at survey altitude over conductive seawater by estimating the distortion of the primary field caused by an in-phase contribution to the primary field from the seawater response (Smith, 2001b).These approximate methods for determining bird position only have an accuracy of a few metres and may therefore be limited.Another method involving the determination of bird position and receiver coil attitude as parameters obtained in a least-squares sense during layered-earth inversion of multi-component datasets (Sattel et al., 2004) has been applied to GEOTEM data to significantly improve the bathymetric accuracy of an AEM bathymetry survey in Torres Strait, located between Australia and Papua New Guinea (Wolfgram & Vrbancich, 2007).

Helicopter time domain AEM
Helicopter TEM systems are similar to the fixed-wing systems except that both the transmitter and receiver loops are located on a framework suspended as a sling load beneath the helicopter, as shown in Figure 2b.The waveforms are similar to those of fixedwing TEM systems, i.e., an on-time period followed by an off-time period (Figure 3).Sattel (2009) has reviewed current helicopter TEM systems.The VTEM (Geotech Ltd.; Witherly, 2004) and HeliGEOTEM/HELITEM (Smith et al., 2009) systems for example are typically used for mineral exploration and the SkyTEM system (Sorensen & Auken, 2004) was developed for hydrogeophysical and environmental applications.The HoistEM (Boyd, 2004) and RepTEM systems have been used for several AEM bathymetry surveys in Australian coastal waters (e.g.Sections 4.2-4.4),together with the SeaTEM system (e.g.Section 4.5) which was developed alongside the RepTEM system8 specifically for bathymetric applications.An example of the SeaTEM waveform is shown schematically in Figure 3d.The HoistEM, RepTEM and SeaTEM systems are central loop systems with the receiver loop and transmitter loops co-axially aligned and lying within the same plane (i.e., nominally no vertical separation between the loops).

AEM footprint
The AEM footprint is a measure of the lateral resolution of an AEM system and was originally studied for frequency-domain AEM in the context of sea ice measurements by Liu & Becker (1990).The AEM footprint, in the inductive limit, was defined as the square area centred beneath the transmitter that contained the induced currents responsible for 90% of the observed secondary magnetic field detected at the receiver (Liu & Becker, 1990).The inductive limit refers to the case of a perfect conductor and/or infinite transmitter frequency, and as such, the induced currents are entirely in-phase with the primary field (i.e.no quadrature component) and the Liu-Becker footprint therefore corresponds to the minimum in-phase footprint of a frequency-domain AEM system.In the case of finite transmitter frequency and earth conductivity, the currents induced in the earth will have a larger spatial extent than that at the inductive limit (Beamish, 2003) and will contain both inphase and quadrature components.Reid & Vrbancich (2004) have compared the inductivelimit footprints of various AEM configurations including time-domain systems in order to analyse the suitability of AEM systems for Antarctic sea-ice thickness measurements.Reid et al. (2006) extended the Liu-Becker footprint calculations (Liu & Becker, 1990;Reid & Vrbancich, 2004) to the case of finite frequency and earth conductivity for HCP and VCX configurations over an infinite horizontal thin sheet and for a homogeneous half-space.This study found that AEM footprint sizes may be several times the Liu-Becker inductive-limit value, with the quadrature footprint approximately half to two-thirds that of the in-phase footprint.
The Liu-Becker footprint is determined by both the transmitter-receiver geometry and the altitude (h).For a dipole-dipole frequency-domain AEM configuration with a transmitterreceiver separation of 6.3 m, the Liu-Becker footprint is 3.73h and 1.35h for HCP and VCX coil geometries.For a central loop system (e.g.HoistEM), the footprint is 3.68h, see Table 2 for a comparison of footprint sizes between several AEM systems.Beamish (2003) also computed the AEM footprint, using a different definition based only on the current system induced in the earth by the transmitter neglecting the contributions the secondary magnetic fields induced by currents in the ground make at the receiver.Beamish determined footprint sizes of between 0.99h and 1.43h for a horizontal magnetic dipole source (transmitter), and between 1.3h and 2.1h for a vertical magnetic dipole source, depending on the transmitter frequency and half-space conductivity.Section 4.4 presents an example of how the EM footprint affects the lateral and vertical resolution using 1D inversion.

AEM bathymetry studies in Australian coastal waters
In 1998, the Australian Hydrographic Service (formerly known as the Royal Australian Navy Hydrographic Service) and the Defence Science & Technology Organisation (DSTO) began investigating the use of AEM as a rapidly-deployable bathymetric mapping technique in Australian waters, 9 complementing the use of lidar.Commercial systems were initially used from 1998 to 2005 for field trials to appraise the accuracy of the AEM bathymetry method based on (i) time-domain fixed-wing AEM using the GEOTEM and QUESTEM systems (Vrbancich et al., 2005a,b;Wolfgram & Vrbancich, 2007), (ii) time-domain helicopter AEM using the HoistEM system (Vrbancich & Fullagar, 2004, 2007b) and (iii) frequencydomain helicopter AEM using the DIGHEM V system (Vrbancich et al., 2000a,b) and DIGHEM_Resistivity system (Vrbancich, 2004).Between 2006 and 2010, field trials were conducted with a prototype SeaTEM system (Vrbancich, 2009), a floating system equivalent to the commercial RepTEM system (Vrbancich et al., 2010) and the SeaTEM system (Vrbancich, 2011;Vrbancich, 2010).The SeaTEM system is essentially a slightly smaller version of RepTEM, developed for DSTO by Geosolutions Pty Ltd. as a research instrument.
The results of the floating AEM system were significant, providing an upper limit to the accuracy expected from AEM systems similar to the RepTEM central loop system.

Sydney Harbour -helicopter frequency-domain AEM (DIGHEM)
A DIGHEM V survey undertaken in 1998 (Vrbancich et al., 2000a) was flown over the width of the entrance to Sydney Harbour including the Sow and Pigs reef located in the centre of the channel.The nominal bird height was 30 m. Two interpretation methods were used to 9 The scope for testing the potential use of AEM for bathymetric mapping is significant, especially within waters affected by turbidity and in the surf zone.Within the Australian Charting Area, the area of waters less than 70 m in depth that are considered to be well surveyed by any method, is estimated to be about 599,000 square kilometres.The areas of waters less than 50 m and between 50 to 70 m in depth that are considered to be poorly surveyed are estimated to be about 531,000 and 867,000 square kilometres respectively.It is also estimated that approximately 30% of the waters within the 70 m depth contour are affected by turbidity.(Australian Hydrographic Service, personal communication, 2010.)estimate the water depths: direct layered-earth (1D) inversion and the conductivity-depth transform (CDT).The former method uses program AEMIE (Fullagar Geophysics Pty Ltd) based on a modified version of a 1D inversion program for horizontal loop EM (Fullagar & Oldenburg, 1984) and the latter method is a rapid approximation method not relying on inversion, to achieve fast processing and interpretation (Macnae et al., 1991;Wolfgram & Karlik, 1995;Fullagar, 1989) using program EMFlow (Macnae et al., 1998).The concept of the conductivity-depth transform is crudely described as follows.For each time in the transient decay (Figure 3c), the conductivity of a homogeneous half-space is computed such that the predicted magnetic field amplitude matches the observed field at that given time.Thus providing the field amplitude at a series of times gives the apparent conductivity10 ( a  ) at each time t.This process associates a set of apparent conductivities with the decay times ( a t   ), and the depths (d) of the "smoke ring" diffusing through the homogeneous ground as a function of time is associated with a set of times ( dt  ), thus the apparent conductivity is associated via time with depth (conductivity-depth transform, a d   ).marked BH, GH and MH in Figures 4a and 5a.The seismic profiles (shown in red) were obtained from an earlier study (Emerson & Phipps, 1969).Another marine seismic study (Harris et al., 2001) confirms that the bedrock reaches the seafloor adjacent to the headland MH at a depth predicted by the AEM data as shown in Figures 4a, 5a, and that the seismic line in Figures 4b, 5b adjacent to the headland BH is incorrect and that the actual bedrock contour in this region is in agreement with that predicted by the AEM data (Figure 5b).

Sydney Harbour -helicopter time-domain AEM (HoistEM)
Sydney Harbour was resurveyed again in 2002 using the HoistEM time-domain helicopter AEM system (Figure 2b).The AEM data were improperly calibrated (Vrbancich & Fullagar, 2004) and was corrected using a procedure that reconciles the measured data with available ground truth at selected points within the survey (Vrbancich & Fullagar, 2007b).An example of a conductivity section resulting from a line that was flown over the Sow and Pigs reef is shown in Figure 6.A two-layer-over-basement model was used for the inversion.
The fitted upper layer conductivity agrees well with the measured seawater conductivity and the seawater-sediment interface accurately follows the known water depth profile (yellow), mostly to within 1 m or better.The maximum depth of investigation was found to be ~ 55 -65 m, just offshore of the harbour entrance.The time-domain equivalent of the skin depth is the diffusion length ( TD where  is the conductivity (S/m) and 0  is the magnetic permeability of free space.Assuming a seawater conductivity of 4 to 5 S/m, TD  corresponds to a depth of 76 to 68 m respectively at the late time ( t ) of 14.4 ms and this represents an estimate of the maximum depth of investigation for this AEM system.Vrbancich & Fullagar (2007a) also showed that this dataset, and hence the AEM technique, has the potential to identify the coarse features of the underlying bedrock topography, as partly revealed in Figure 6.shown Figure 6, to a gridded surface (i.e. by gridding the depth profiles).The upper image was obtained by gridding the known water depths.For the purpose of this comparison, individual known bathymetry grids from several high-spatial density sonar surveys and airborne laser depth soundings were combined into a single grid and sampled at the HoistEM measurement locations (fiducial point coordinates) to produce bathymetric data at the relatively low spatial density of the AEM data, e.g., the black and yellow curves in Figure 6.Gridding the series of yellow curves across all the survey lines yields the upper image in Figure 7. Beach and this section and the coastal area adjacent to the beach follows the seafloor gradient accurately (compare with black and grey profiles obtained from sonar data) to a water depth of about 25 to 30 m.The peaks at ~ 301400 and 302300 m (E) arise from reef structure, identifying a cross-section of the "zig-zag" nature of palaeochannels incised into the bedrock, as shown by the troughs located at 301800 and 301300 m (E) (Vrbancich et al., 2011).The AEM survey was flown using 30 lines with 100 m line spacing.A colour-shaded surface grid obtained by gridding all the 30 profiles is shown in Figure 9 and can be compared directly with an equivalent grid of the single-beam sonar data, shown in Figure 10.All of the significant features and most minor features are accurately identified in the AEM   bathymetry map which took approximately 2 hours to survey and about 1 hour to process the data and run the inversions to produce a georeferenced map.The SeaTEM AEM system is experimental.In some regions, as shown in Figure 9, the interface depths display an oscillatory profile in some areas which degrades the accuracy of the derived water depths.
Intensive tests conducted after this survey have shown that these anomalies may be caused by a high frequency noise source which cannot be removed by the usual data processing procedures.The source of this noise is currently being investigated.

Yatala Shoals -helicopter time-domain AEM (HoisTEM) -AEM footprint features
Yatala Shoals (~ 35.7º S, 138.2ºE) is located in Backstairs Passage, South Australia, between the Fleurieu Peninsula and Kangaroo Island.This area is interesting because it contains a series of ridges, which fan out, increasing in separation and decreasing in height above the seafloor.This area was surveyed in 2003 using the HoistEM system, to examine the effect of the EM footprint on the lateral resolution, for resolving peak separation and height.
Figure 11 shows the results of layered-earth inversion for two profiles, separated by about 300 m.The main feature is a ridge rising approximately 18 m from the seafloor at ~ 244700 m (E) (Figure 11a) which gradually deepens and splits into two ridges (244390 and 244670 m (E), Figure 11b).A series of narrow ridges and troughs flank the eastern side of this main ridge.The impact of the AEM footprint is evident as a smoothing and underestimation of peak heights.Referring to the AEM bathymetry profiles (red), the narrow trough (~ 100 m gap) to the west of the main ridge in Figure 11a between 244400 and 244700 m (E) is unresolved, yet as this ridge splits into two ridges with a trough wider than 300 m (Figure ), the height and width of the western ridge (~ 244390 m (E)) is well resolved, and the secondary peak at 244670 m (E) is clearly resolved, yet its width is over determined and the height of the ridge is noticeably underestimated by about 3 m.The ridges that flank the eastern side of the main ridge are relatively narrow (~ 100 to 150 m) and are smaller than the central loop EM footprint, thus whilst the peaks are identified, their elevations above the seafloor are all underestimated by several metres.Another factor contributing to the apparent smoothing of the peak structure is the effect of assuming a 1D (layered-earth) model for inversion.2D/3D modelling and inversion would be more appropriate in this case, however the routine use of generalised 3D modelling and inversion methods has not been realised, furthermore, the application of existing methods to an entire AEM survey dataset is impractical because of the required computer processing time.Cox et al. (2010) have recently introduced a robust 3D inversion scheme based on a moving AEM footprint that would enable the effects of the 1D approximation (i.e.assuming a layered-earth) to be fully investigated.The extent and depth of the palaeochannel in this area was found to be in very good agreement with a depth-to-bedrock map derived from marine seismic data (Vrbancich et al., 2011b).

Discussion and conclusion
The use of AEM methods, traditionally applied to mineral exploration, can be applied to the measurement of water depths and seawater and seabed conductivity in shallow coastal waters that may be turbid and lie within the surf zone, thereby limiting the application of lidar techniques in these waters.The pioneering work in this area was carried out in Canada and the USA, with most of this work published in the 1980s and early 1990s11 .This paper presents some of the findings of AEM bathymetry studies in Australian coastal waters carried out between 1998 and 2010 using helicopter frequency-and time-domain systems.
The AEM footprint (and hence the lateral resolution) is significantly greater than the lidar footprint, and this limits the AEM method for bathymetric mapping where International Hydrographic Organisation (IHO) standards are required.However sub-metre vertical water depth resolution can be obtained using AEM methods in shallow waters (within a 10 to 20 m depth range).Presently, the AEM bathymetry method is a reconnaissance technique that can be rapidly deployed in remote areas via helicopter, for fast estimation of coastal water depths, including waters in the surf zone and turbid waters, to depths of about 30 to 40 m, and for identifying areas of exposed reef and for estimating the coarse features of the underlying bedrock topography.One of the most important issues is instrument calibration, affecting both environmental and exploration applications of AEM.Data obtained from (i) measurements of seawater depth and conductivity, (ii) sediment conductivity from bore hole samples and (iii) sediment depths estimated from marine seismic data, all acquired from a suitable trial site, can be combined to provide ground-truth data to establish a suitable geo-electrical model, for checking instrument calibration errors by comparing derived depths and conductivities with "known" depths and conductivities.Improved instrument calibration will lead to more robust estimates of seawater depth, seabed properties, sediment thickness and bedrock topography as well as improved depth accuracy and deeper depths of investigation.
Apart from improving AEM instrument calibration, reducing noise sources and including sensors to accurately track bird motion, future work will also involve software enhancements in program AMITY and associated software (Fullagar Geophysics Pty Ltd) to enable geologically-constrained 1D time-domain inversion, and 1D extremal inversion which enables models to be constructed with maximal and minimal characteristics (e.g.depth and conductivity bounds) which fit the observed data to within acceptable levels.
Where warranted, future studies will also include the use of 3D EM modelling and inversion methods for interpretation of AEM data.

Fig. 1 .
Fig. 1.Schematic diagram of the fixed-wing time-domain GEOTEM (Fugro Airborne Surveys Pty Ltd.) AEM system.The transmitter loop is mounted around the aircraft and the receiver loop is trailed behind and below the aircraft.

Fig. 2 .
Fig. 2. (a): the frequency-domain helicopter DIGHEM V AEM system (~ 8 m length) during survey over Sydney Harbour(Vrbancich, et al., 2000a,b).The smaller bird between the helicopter and the DIGHEM bird is a magnetometer bird; (b): The HoistEM time-domain helicopter AEM bird, located over the Sow and Pigs reef during a survey of Sydney Harbour(Vrbancich & Fullagar, 2004, 2007b).The transmitter loop (~ 22 m diameter) is attached to the extremities of the poles and the multi-turn receiver loop is located on the same plane, at the centre of the system.

Fig. 4 .
Fig. 4. AEMI (layered-earth inversion) conductivity sections (S/m) with profiles of depth to bedrock from seismic surveys (red), echo sounding water depths (black) and Geoterrex-DIGHEM proprietary inversion software (white).(a): this line skirts the coastline and lies closest to the headlands, marked BH, CH, GH and MH; (b) and (c) progress further away from the coastline; line spacing is nominally 50 m.Source: Vrbancich et.al., 2000b.

Fig. 6 .
Fig. 6.Conductivity-depth sections from 1D inversion using corrected data to account for imperfect AEM instrument calibration.Accurate bathymetry profiles (obtained from combined multi-beam sonar and lidar data) are shown: (i) tide corrected (black) referred to a given datum; (ii) actual water depth at time of survey (yellow), which includes a tide height of 1.4 m.The Sow and Pigs reef is shown as a pinnacle at ~ 339900 m (E).The colour scale has units of S/m.Measured seawater conductivity is ~ 4.7 S/m.The green section represents sediment and the blue sections represent underlying resistive basement (bedrock).Source: Vrbancich & Fullagar, 2007b.

Figure 7
Figure7compares 3D images of the seafloor topography (defined by the water depth); the lower image is derived from inversion of corrected AEM data and was obtained by interpolating the water depth profiles defined by the seawater-sediment interface depth, as

Figure 8
Figure8shows a representative example of a profile (line L1185, red) of interpreted water depths obtained from layered-earth inversion of SeaTEM data, from a survey flown over Crookhaven Bight(~ 35.0º S, 150.8ºE)  in November 2009, adjacent to Jervis Bay and located approximately 150 km south of Sydney.The western section of the profile joins Currarong Beach and this section and the coastal area adjacent to the beach follows the seafloor gradient accurately (compare with black and grey profiles obtained from sonar data) to a water depth of about 25 to 30 m.The peaks at ~ 301400 and 302300 m (E) arise from reef structure, identifying a cross-section of the "zig-zag" nature of palaeochannels incised into the bedrock, as shown by the troughs located at 301800 and 301300 m (E)(Vrbancich et al., 2011).

Fig. 7 .
Fig. 7. Three-dimensional images of the seafloor topography based on pre-existing sonarlidar data (top) and on inversion of corrected HoistEM data (bottom).Inverted seawater depths were not smoothed prior to gridding.The images are vertically aligned and are depicted with the same vertical exaggeration and colour scale.The surface is coloured according to water depth: red (0-4 m); orange (< 7m); yellow (<10 m); pale green (< 16 m); aqua (< 23 m) and dark blue (< 32 m).The "white" arrowhead (top) marks the location of the Sow and Pigs reef.Source: Vrbancich & Fullagar, 2007b.

Fig. 8 .
Fig. 8. Seabed level relative to Australian Height Datum (AHD) for survey line L1185; red: AEM bathymetry; black: Australian Hydrographic Service single-beam sonar data; grey: single-beam sonar data recorded during marine seismic survey (Vrbancich et al., 2011a).The location of this profile is shown in Figures 9 and 10.
Fig. 8. Seabed level relative to Australian Height Datum (AHD) for survey line L1185; red: AEM bathymetry; black: Australian Hydrographic Service single-beam sonar data; grey: single-beam sonar data recorded during marine seismic survey (Vrbancich et al., 2011a).The location of this profile is shown in Figures 9 and 10.

Fig. 9 .
Fig. 9. Bathymetry of Crookhaven Bight survey area derived from AEM (SeaTEM) data.The bathymetry profile for line L1185 is shown in Figure 8 (red).The colour scale refers to the water level (m) relative to Australian Height Datum.

Fig. 10 .
Fig. 10.Bathymetry of Crookhaven Bight survey area derived from sonar (Australian Hydrographic Service) data.The bathymetry profile for line L1185 is shown in Figure 8 (black).The colour scale is the same as shown in Figure 9.

Fig. 12 .
Fig. 12. Bathymetry maps of an area adjacent to Palm Beach, Broken Bay: (a), sonar; (b) AEM.Both maps use the same colour scale to show the seabed level relative to the Australian Height Datum.

Table 1 .
Frequencies and associated skin depths (m) for the DIGHEM V and DIGHEM-Resistivity AEM birds.Skin depth  (m) assumes a typical seawater conductivity of 4 S/m.Depth of investigation is equivalent to approximately 2 (m).

Table 2 .
Reid & Vrbancich (2004)mit footprint sizes (footprint to transmitter height ratios) for several common AEM systems.Tx: transmitter, Rx: receiver.For TEM systems, the geometry is specified by the directions of the Tx and Rx dipole axes: Tx(z)-Rx(z) denotes a vertical-axis Tx and vertical-axis Rx, Tx(z)-Rx(x) denotes a vertical axis Tx and a horizontal-axis (inline) Rx.For the central loop helicopter TEM configuration (e.g.SkyTEM, HoisTEM), the Tx loop was a finite horizontal loop (10 m x 10 m) and the Rx was a vertical axis point dipole.Adapted from Table1,Reid & Vrbancich (2004).