Andrew Long of Petroleum GeoServices provides a rough guide to the science behind broadband seismic and why it is transforming the possibilities for acquiring high resolution images of the subsurface using towed streamer marine seismic acquisition technology.
In any towed streamer seismic acquisition project there are three considerations regarding the bandwidth of signal from the earth available in the final seismic image product: 1) the frequency bandwidth propagated into the earth from the source array; 2) the frequency bandwidth recovered from the earth in the recorded data; and 3) the frequency bandwidth preserved throughout all processing and imaging steps.
Seismic signals are described in classical terms of amplitude, phase, and frequency content. Each component must be faithfully preserved in order to accurately interpret geological structure and stratigraphy, and to accurately predict lithology and fluid distribution during reservoir characterization. The latter pursuit benefi ts in particular from very low frequency amplitudes being recovered from the earth. A generic defi nition of ‘broadband seismic’ thus describes an acquisition and processing system with source and receivers which enhances and preserves the bandwidth at both low and high frequencies in a prestack amplitude and phasecompliant manner so that subsequent processing and interpretation can utilize all the information contained in the signal from the earth.
‘Ghosts’
Unwanted reflections from the freesurface of the ocean continuously interfere in a constructive and destructive manner with the seismic wavefield propagated into the earth from a source array. The source wavefield reflected from the surface (the ‘source ghost’) is a timedelayed and opposite polarity version of the source wavefield propagated directly from the source array into the earth, and the two wavefields propagate together in a coupled manner. The net effect is that the frequency bandwidth propagated into the earth contains signifi cant notches at periodic frequencies, and the notch frequencies are a function of both source depth and emission angle (measured with respect to vertical).
Similarly, the receivers (along each streamer) record two versions of the seismic wavefield scattered back from the earth, coupled together and interfering in a continuously constructive and destructive manner. The wavefield reflected downwards from the freesurface of the ocean (the ‘receiver ghost’) is referred to as the ‘downgoing’ wavefield, and is a timedelayed and opposite polarity version of the ‘upgoing’ wavefield. The wavefield recorded with conventional hydrophoneonly streamers is a scalar measurement of pressure; the ‘total pressure’, which is the sum of the upgoing and downgoing pressure wavefields. The recorded total pressure wavefield contains signifi cant notches at periodic frequencies, and the notch frequencies are a function of both receiver depth and emergence angle (measured with respect to vertical). So collectively, conventional seismic data contains frequency notches related to both source ghost and receiver ghost effects. These effects notably penalize the low and high frequency content in seismic data, resulting in a limited frequency bandwidth being recovered from the earth.
Physics describes how any pressure wavefield can also be defi ned in terms of the derivative of pressure normal to the wavefront; measured in units of particle velocity. Figure 1 (shown overleaf) illustrates how the receiver ghost notch frequencies are complementary for pressure and particle velocity wavefields, and how the notch frequencies change as a function of emergence angle. There is usually no usable information in the vicinity of the spectral notches, so any processingbased solution to recover information in these parts of the spectrum must be based on reconstructing the data that have not been recovered from other parts of the data with higher signaltonoise (S/N) content.
Figure 1. illustrates how the pressure (blue) and particle velocity (red) amplitude spectra are complementary when measured at the same depth and location (‘collocated’). Periodic notches in both spectra are related to the receiver depth and the angle of emergence of the seismic wavefield. As the emergence angle increases (vertical propagation means zero emergence angle), the notches move to higher frequencies.
Figure 2. The image on the left is the result of seismic inversion applied to conventional seismic data containing both source and receiver ghost effects. The color scale represents ‘P impedance’: the product of compressional velocity and density. In contrast, the right image represents the ghostfree result provided from PGS GeoSource and dualsensor GeoStreamer technologies. Note the improvements in resolution on the right.
Traditionally, this involves simple 1D deconvolution of the data using a deterministic assumption that the sea surface is perfectly flat, streamer depth is constant, and the earth and water column is homogeneous. Inevitably, such methods are bound to fail as the various assumptions are increasingly violated.
Several acquisitionbased methods have emerged that do not seek to mitigate the presence of the ghost notches. They record information with different ghost characteristics such that, when all the data are combined, there is good S/N at a wider range of frequencies:
Each methodology requires rather exhaustive explanation to describe its implementation, but the common element is that a reflection wavefield approximating the true upgoing pressure reflection is derived in processing, and the effects of the receiver ghost are removed.
The past couple of years have also seen a variety of source array approaches that deploy source elements at two or more different depths, and processing is able to reduce or remove source ghost effects. It is also noted that a family of processingbased methods have also emerged in recent years that attempt to reduce or remove source and/or receiver ghost effects from conventionally acquired seismic data. Each makes a series of assumptions, but results can be favorable in certain scenarios. Figure 2 shows a comparison of conventional source and receiver seismic data vs ghostfree seismic data.
Using ghostfree data
As demonstrated in Figure 2, removing the effects of the source and receiver ghosts significantly improve the frequency bandwidth recovered from the earth, and facilitate high resolution interpretation. Ghostfree data is in fact a prerequisite for many processing algorithms and inversion schemes. Overall, each acquisition and processing ‘solution’ to mitigating ghost effects and increasing recoverable frequency bandwidth is based upon several assumptions. In ‘optimal’ survey conditions and in locations with naturally high S/N seismic images the various ‘broadband’ results may be quite comparable in terms of image quality. But the industry is still in the process of understanding the penalties for reservoir characterization and image quality as various assumptions in each methodology are violated in the acquisition and survey environmental parameters, and in terms of various geological settings and styles.
The most robust broadband seismic solutions are based on an acquisition platform, but even then the industry is still learning how to best process such data. OE

Overunder hydrophoneonly streamers. l Variabledepth hydrophoneonly streamers.

Dualsensor streamers (include vertical component particle velocity sensors).

Multicomponent streamers(include vertical and crossline component particle velocity sensors).
