We describe a method to remove the air wave from shallow marine transient EM data. At a long enough offset the air wave dominates the measurement and the response of the earth is negligible. Thus a long-offset measurement can be used to estimate the shape of the air wave which is the same at short offsets, provided the water is shallow. We remove the air wave by filtering. We demonstrate the method on both synthetic and real data.
In marine Controlled-Source ElectroMagnetic (CSEM) surveying there are essentially three propagation paths from the current bipole source to the receiver, as depicted in Figure 1: directly through the ground, directly through the water, and via the water to the air and back down to the receiver. Since the wavefield is continuous everywhere, these three paths are not independent. Energy that travels by the third path is known as the ‘air wave’. Propagation from the source to the sea surface is attenuated because water, especially sea water, is an excellent conductor of electricity. At the sea surface there is a pure inductive effect that can be regarded as instantaneous: the flow of current produces a magnetic field which induces an electric potential in the water, causing current to flow. If the depth of the water is zero, the air wave arrives instantaneously at the receiver. Conventional CSEM surveying uses a continuous signal from the source, for instance a square wave, that contains several frequencies, and the received data are generally analysed in the frequency domain. Inversion of the data by iterative forward modelling can include the water layer. In shallow water, because the response is dominated by the air wave, the inversion is most sensitive to changes in the water layer parameters and least sensitive to changes in sub-sea parameters. This has been known for a number of years and considerable effort has been devoted to the removal of the air wave. (Amundsen, 2003; McGregor and Sinha, 2004; Rosten and Amundsen, 2004; Lu et al., 2005; Amundsen and Holvik, 2006; Madsen, 2006; Weidelt, 2006).
The target reservoir is typically at a depth 1-4 km below the sea floor and offsets r in the range 2-4 times the target depth are normally required to image the target. Continental shelf water depths up to 200 m are small compared with these offsets. The first step is to recover the system response s(t) . If both v(t) and i(t) are measured (Wright et al., 2005), this is a straightforward deconvolution problem. The response s(t) is the sum of the three arrivals described above. The resistivity of the earth is normally greater than the resistivity of the water. The propagation velocity is roughly proportional to the resistivity, so the earth response arrives before the water arrival, and if the water depth is small, this arrival can be neglected within the time interval of interest. For water of zero depth the air wave is an impulse.
