Grounded Sources
Presentation Highlight
Grounded sources is the fourth and most involved topic we cover in the DISC presentations. It combines physical phenomena from both direct current resistivity and inductive source electromagnetics (EM). In a grounded source EM experiment, we position electrodes on the surface (or in the ocean in the case of marine applications) and inject a time-varying current.
Grounded source methods can be applied both in marine settings and on-land. In the context of marine electromagnetics, de-risking hydrocarbon reservoirs is arguably the most common application. Seismic is the workhorse geophysical method for oil and gas exploration. In marine settings, it can be used to identify potentially prospective hydrocarbon-bearing reservoirs. However, it has minimal sensitivity to hydrocarbon saturation — from seismic velocities alone, we cannot tell the difference between a low-saturation reservoir consisting of fizz-gas (~80% water and ~20% gas), and a more prospective, saturated reservoir (eg. >60%). Resistivity however, is more diagnostic as a physical property for indicating hydrocarbon saturation, and thus EM is a technique of interest. The Barents Sea Case History is one example demonstrating the application of Marine EM for de-risking a hydrocarbon drilling project in the North Sea. Newer on the scene is the application of marine EM for locating seafloor massive sulfides as well as methane hydrates. On land, grounded source EM is used in mining applications and hydrocarbon exploration & monitoring.
Grounded systems make use of both DC and EM inductive behaviours, thus we can build up our understanding of grounded source EM using concepts from both DC and EM induction. We have two current electrodes, an injection electrode and a return electrode, as well as a wire. Since we are putting a time-varying current through the wire, we will produce time-varying magnetic fields. To piece together the physical processes, we will make use of our understanding of charges, and how they build up at conductivity interfaces from DC, and of currents, and how they are concentrated in conductors from inductive EM. For a refresher, take a look at the DC resistivity presentation and the Inductive Sources presentation
As a first model, we will consider an electric dipole in a whole space. An electric dipole is essentially a generator connected to two prongs that are stuck into the earth. At steady state, we have the DC solution for an electrostatic dipole in a wholespace. Now what happens if we turn the generator off?
Very shortly after shut-off (eg. 1e-4 ms), the currents looks like that of a dipole. To gauge how signals propagate through time, we use the diffusion distance which is an indication of how far EM energy has propagated through the conductive earth at any point in time; it is analogous to skin depth in the frequency domain. At early times, the diffusion distance is small, as most energy is concentrated near the source location. As we look later in time, the concentration of currents diffuses outwards (also notice that in the figure above, the colourbars are changing through time), this is also reflected in the increase in diffusion distance with time. Initially the currents were focussed at the dipole, after shutoff, the peak at dipole is gradually reduced.
If we look in the frequency domain to see what happens to our dipole, we can make a comparison with TDEM. The additional complication that working in the frequency introduces is the partitioning of a signal into the real and imaginary components. If you take reciprocal of time (giving a frequency), and compare the FDEM response at that frequency, we can see that the behaviour of the fields is not so dis-similar to what we observe at the associated time in the TDEM. At early times and high frequencies, the currents are concentrated near the source. As we move to later times and lower frequencies, they diffuse outwards. Similar to diffusion distance in the time-domain, skin depth provides an indication of how far signals have propagated.
Lets make things more interesting and use a wire, it could be straight or laid out as a horse-shoe. We put a current through the wire: at the A electrode, we inject current into the ground and at the B electrode we extract current. We turn the current on and set up steady-state, DC currents in the earth (as shown in the figure on the right above). Now for the interesting part: What happens when we turn the current off?
To answer this, we break down the currents into two parts: (1) the currents in wire and (2) the currents in the ground.
Lets start by considering the currents in the wire (the inductive part). The moment the generator is turned off, a large magnetic flux is generated due to the wire. Now the earth opposes change in flux, so immediately after shut-off that current is transferred into the ground; this is often referred to as an image current. The current, now in the ground, flows in the same direction as the current that was in the wire. Then, as soon as you put currents into a conductive earth and don’t have a forcing function, they will start to diffuse downwards and outwards through time.
Next, lets consider the currents in the ground (the galvanic part). This is a bit simpler. We start with DC currents in the earth. When we shut off the generator, we remove the forcing function driving the currents and the currents in the ground will simply diffuse downwards and outwards.
The net response due to a current shut-off in a grounded wire, is then just the combination of these two effects. In the figure on the right above, you can see that at early times, the currents from the wire are immediately transferred to the earth. Both the image current and the ground currents then diffuse down and out.
Now lets suppose that we put a target in — some kind of a conductive target…
We consider a conductive block in a halfspace. Prior to shut-off, we have set up steady-state galvanic currents. At this point, it is a DC resistivity experiment and charges build up at the interfaces and currents to are channeled through the conductive target. Then we shut the current off. In doing so, we create a time varying magnetic field that will impinge on the block. A time varying magnetic flux through a conductive target will generate currents that oppose that change in flux (according to Lenz’ Law). Thus, we generate vortex currents in the block. These are distinct from the galvanic currents in the DC problem because no charges are being built up.
As we look later in time, we see that the currents are primarily going through the block — they are galvanic currents being channeled through the block. We have shifted from primarily seeing vortex currents to observing galvanic currents. In summary, there is a combination of galvanic and inductive effects: prior to shut off, galvanic effects are the only contribution. At early times after shut-off, the induced vortex currents are the main effect in the conductive target. At later times, the vortex currents are gone and we are again seeing galvanic current channeling through the conductive block.
This gives us an idea of where the currents are in the earth for this experiment. Now suppose that we want to go out and measure data on the surface. We can measure either of the two horizontal components of the electric field (x or y… z is not well defined at the surface) and any of the three components of the magnetic field. The budget is limited and you want to maximize your time. Which fields and components of those fields should you measure? why?
Lets start by considering the electric field. We can’t practically measure the vertical component of the electric field, so we are left considering the horizontal components. If the x-direction is parallel to the wire and the y-direction is perpendicular to the wire, which should we measure? To answer this, we need to consider coupling, both for the galvanic contribution and the vortex currents. The galvanic currents are mainly flowing in the plane of the transmitter, as are the vortex currents. The electric fields are in the same direction as the currents, so our best coupling is in that same direction; thus we want to measure the x-component. Examining the data plots in the left panel of the figure above, we see that there is a significant difference between the x-component of the electric field measured over a halfspace and that measured when the block is present. This is true at early, mid, and late times as we might expect since the conductive block alters the galvanic currents and gives us vortex currents; the combination of these effects spans early, mid and late times. Thus, the x-component of the electric field is sensitive to the presence of the block for this set-up.
Now to the magnetic flux, b; we can measure any of the 3 spatial components, which should we choose? The magnetic fields are rotational around the currents (think: right-hand-rule). Directly above the wire, the magnetic fields due to both the galvanic and vortex currents are going in and out of the page; they are oriented in the y-direction. If we take a step away from the wire, we also have a vertical component of the magnetic field and therefore will also want to measure the vertical component of the magnetic field. Examining the data plots in the center panel (showing the y-component) and right panel (showing the z-component) of the figure above, we see that both the y and z components of the magnetic fields show differences between the halfspace and the model with the block, particularly at the mid and later times.
Now what happens if our target is instead a resistor?
Lets consider the same two phenomena we were discussing for the conductive target: a galvanic contribution and an inductive contribution. First consider the galvanic contribution — in this case, the currents divert around the target (just as in DC Resistivity) and the charges will be opposite to what we observed for the conductive target. What about the vortex currents? Although time-varying magnetic fields produce time-varying electric fields everywhere, currents are only induced in conductors (as per Ohm’s law), thus for a resistive target, there are no vortex currents induced; for a resistor, galvanic effects dominate.
Given that, which fields and which components do we want to measure?
Since galvanic effects dominate for a resistor, we expect that the build up of charges is the main effect we want to observe, and again, due to the coupling of the transmitter and target, we expect the x-component of the electric field to be sensitive to the resistor. Looking at the left panel of the figure above, we can see that there are differences in the electric fields measured over a half-space and those measured over a half-space with a resistive block. The magnetic fields are quite similar for the half-space and the model with the resistor; the magnetic fields are not particularly sensitive to the presence of a resistive target. This is why, for de-risking hydrocarbon reservoirs using marine EM, where the target is a resistor, electric fields are what is measured.
In summary, for the setup we are looking at, if we are looking for a conductive target, the inline electric and the vertical and tangential magnetic fields are all valuable data to collect whereas if the block is a resistor, the inline electric field is most sensitive to the target, the magnetic fields are hardly affected by its presence. Now, in general, geologic structures are more complex than a block in a half-space that is happily centred with our source and the fields can get complicated quickly. Thus, forward simulations are necessary for examining which components of the fields are sensitive to the target of interest.
Deccan Traps
Case history highlight
The Deccan Traps are basalt flood plains in India. They were of very public interest a number of years ago because there were questions about the extinction of the dinosaurs. There were questions about whether an asteroid impact or extensive volcanism was responsible for killing the dinosaurs; both the extensive eruptions that led to the Deccan Traps and the Chicxulub impact happened almost contemporaneously ~66 million years ago. Eventually, it was shown that the asteroid impact was responsible for the extinction of the dinosaurs.
This is an older study, conducted 10 years ago now, so numerical techniques have since advanced, however, this example demonstrates aspects of quality survey-design and is a success-story in the application of grounded source EM for answering a geologic question. A write-up of the case history is available on em.geosci.
Characterizing the geology beneath the Deccan traps is of interest as there are sediments beneath which are potentially hydrocarbon-bearing. Knowing the thickness of the sediment is important for understanding prospectivity. The area of interest for this study is the Saurashtra basin in Eastern India. Seismic is often the workhorse geophysical method for hydrocarbon applications, however basalts have a very high seismic velocity and the base of the basalts is a significant seismic reflector thus imaging beneath it is challenging using seismic. The basalt structure is also generally complicated resulting in significant ringing effects. Resistivity however, is potentially diagnostic in this setting. The Deccan Trap basalts are resistive (150Ωm — 600Ωm) and the sediments beneath are relatively conductive in comparison (30Ωm — 100Ωm); beneath the sediments is a very resistive basement.
As an initial experiment, a DC resistivity survey was run along a transect running approximately east-west. This experiment showed that electrical methods could delineate the thickness of the sediments and motivated further investigation into the the use of electromagnetic methods for estimating the thickness of the sediments and how the thickness varies. In particular, one hypothesis suggested that the sediments may thicken towards the south.
The next survey considered is a Long Offset Time Domain EM (LOTEM) experiment. This is a grounded source experiment in which a long transmitter wire is connected to the ground, and electric and magnetic field receivers are positioned at some distance away from the transmitter (see the left image above). The transmitter is kept stationary and the receivers are moved to increase coverage.
From the point of view of survey design, we want to see what we can expect of our data before heading out to the field. Can we expect to see differences in our data if the thickness of the trap basalt varies? And our prime goal: do we expect our data to be sensitive to the thickness of the sediments? To address these questions, the authors simulate data over a layered model in which the thickness of the units of interest are varied. They showed that there are significant differences in the simulated data for different trap-basalt thickness as well as for different sediment thickness.
Data were collected in the late 1980’s along 8 survey lines. The data were inverted using 1D parametric inversions and the results stitched together to create profiles of each survey line. In the image on the right above, 4 cross sections are plotted. The trap basalts are 80–100m thick across most of the surveyed area. As for the sediments, the inversion results are contrary to the initial prediction — they show the sediments thinning, rather than thickening, towards the south.
In an EM inversion, it can be difficult to tease apart the conductivity and the thickness of a conductive layer — if we compare data collected over a model having a thin layer with conductivity, x, and thickness, t, with data from a model with a thinner (0.5 t) more conductive layer (2 x), those data will be quite similar. This is an example of non-uniqueness (or equivalence) with respect to the conductance (the product of the conductivity and thickness of a unit). To formulate an interpretation of the sediment thickness, the authors compute the conductance of the sediment unit found in the inversion result and interpolate this to create the map shown in the figure on the left above.
The conductance map shows that the region of thickest sediment appears to be along line L (the red region). Based on this, a well was drilled in this area. From the 1D inversions, the thickness of the sediment was estimated to be ~1.5km. This estimate was compared to the drilling results and showed good agreement (figure on the right above).
This example shows the application of EM at large scales (100’s of kilometres); once you understand a concept of EM at one scale, you can scale it up or down, and the same principles apply.
