Method of and Apparatus for Exploring a Region Below a Surface of the Earth

The present invention relates to a method of exploring a region below a surface of the Earth. The present invention also relates to an apparatus for performing such a method, a program for performing data processing of such a method, a computer programmed by such a program, a computer-readable medium containing such a program and transmission across a network of such a program. Such a method may be used, for example, to search for new sources of hydrocarbons and to monitor known hydrocarbon reserves.

There are various known techniques for exploring the structure and/or composition of regions below the surface of the earth, including below the sea bed. Some of these techniques are "active" in that they require the use of a specific controlled source of energy to probe the region and then measure the effect of the region on the transmitted energy. A typical example of such an active technique is the well-known seismic exploration technique, in which one or more sources produce impulsive or swept- frequency pressure waves which penetrate the earth and the returned pressure variations are measured by geophones or hydrophones.

Other techniques are of the "passive" type in that they do not require energy to be produced by a source but, instead, make use of already-existing energy to provide information about the structure or composition below the surface of the earth. One such technique is known as hydrocarbon microtremor analysis (Hymas), for example as disclosed in Dangel et al, 2001 , "Phenomenology of tremor-like signals observed over hydrocarbon reservoirs (just related and higher frequencies)". Hymas attempts to measure a continuous signal or tremor in a frequency range between 1 and 10 Hz over the reservoir (not "outside" or "beside" it). As illustrated in Figure 1 of the accompanying drawings, sensors such as 1 and 2 embedded in the surface 3 of the earth measure such tremors. The sensors 1 , 2 are in the form of extremely sensitive seismometers having a sufficiently broad bandwidth to be sensitive to the frequency range of interest. The seismometers are typically arranged as a two-dimensional array or grid with the spacing between seismometers typically being from a few hundred

metres to a kilometre, depending on the spatial resolution required. The outputs of the seismometers are recorded for a period of typically 24 hours. It is not necessary for the recordings to be made simultaneously and fewer seismometers may be used by changing their locations after each recording period.

This technique is based on the assumption that a hydrocarbon reservoir 4 interacts with ambient seismic waves to modify the waves in a measurable way. Figures 2 and 3 of the accompanying drawings illustrate this. The seismometer 1 is located such that there is no hydrocarbon reservoir in the region of the earth below it. Figure 2 illustrates the result of analysing the recorded data from the sensor over a recording period of typically 24 hours. The resulting data are plotted as a frequency spectrum of amplitude in arbitrary units against frequency in Hertz (Hz). The seismometer 2 is located above the hydrocarbon reservoir 4 and the spectrum obtained from the recorded data is shown in Figure 3. It is believed that the presence of significantly increased energy in the frequency band around 3 Hz results from the modified seismic waves 5 produced by the hydrocarbon reservoir 4 and so provides a direct indication of the presence of the hydrocarbon reservoir.

A problem with the Hymas technique is that it is based on the detection and recording of very weak seismic tremor signals. However, human activity provides relatively high amplitude noise in the frequency range investigated in the Hymas technique. For example, noise resulting from vehicles, pumps and drilling occurs in this frequency range and may make the Hymas technique unusable in many situations.

Another known passive technique is referred to as ambient noise tomography (ANT). This technique is based on the use of conventional seismometers which have been recording very low frequency ambient noise for very long periods of time, for example of the order of years. The frequency range of interest is below 0.5 Hz and mainly below 0.05Hz. Although the seismometers may be spaced apart by hundreds of metres for special cases, they are typically spaced apart by hundreds of kilometres and even globally around the earth. It is necessary for recordings to be made simultaneously and this reveals subsurface information between the locations of pairs of seismometers. An example of this technique is disclosed in Bensen et al, 2007,

"Processing seismic ambient noise data to obtain reliable broad-band surface wave dispersion measurements".

The data from such seismometer arrangements are processed so as to extract the Green Function by cross-correlation of two simultaneous recordings. The results for each frequency are then used subjected to a dispersion analysis followed by a tomographic inversion so as to obtain a laminar image. The resolution is such as to provide structural information about the earth on the continental scale. Thus, this technique is generally not of particular interest when looking at smaller regions, for example, when exploring for or monitoring hydrocarbons, because of the insufficient spatial resolution.

Russian Patent no. 2271554C1 discloses a technique for recording and analyzing ambient noise in a frequency range below 15Hz. In a first phase of this technique, ambient noise is simultaneously recorded ("synchronised") by means of at least three sensors which are sensitive to the vertical component of the ambient noise signal. The data are then analyzed to estimate dispersion curves within the area being investigated.

In the next phase, ambient noise is recorded using at least two synchronised sensors sensitive to the vertical component. One of the sensors is used as a permanent station at the centre of the area being surveyed during the whole of the survey. The or each other sensor is moved to different locations within the area being surveyed so as to record the vertical component of the ambient noise at each of the locations for an unspecified time period. The recordings obtained by the non-fixed sensors are then calibrated by calculating the differences between the power spectra of the permanently located sensor and the "moving" sensors. From the estimated dispersion curve, a depth is calculated, in an unspecified way, for each frequency. Maps are then drawn for each frequency or depth of the power spectra differences. Because the sensors are sensitive only to the vertical component of the measured ambient noise, this technique is limited to Rayleigh waves.

According to a first aspect of the invention, there is provided a method of exploring a region below a surface of the Earth, comprising: using a plurality of sensors at a

plurality of locations to obtain seismic data by recording ambient seismic interface waves, each sensor being sensitive to three substantially orthogonal components of the waves in a frequency range whose lower limit is greater than or equal to OHz and whose upper limit is less than or equal to substantially 1 Hz; and processing the data to obtain a measure of the energy in a frequency band within the frequency range.

Each of the sensors may be sensitive to two orthogonal horizontal components and one vertical component.

The method may comprise rotating the components so that one thereof is aligned in a desired direction. The desired direction may be a direction of arrival of the waves.

According to a second aspect of the invention, there is provided a method of exploring a region below a surface of the Earth, comprising; using a plurality of sensors, each of which is located in turn at a plurality of locations, to obtain seismic data by recording ambient seismic interface waves in a frequency range whose lower limit is greater than or equal to OHz and whose upper limit is less than or equal to substantially 1 Hz; and processing the data to obtain a measure of the energy in a frequency band within the frequency range.

The method may comprise locating the sensors at a plurality of different combinations of the locations in turn and making at least partially simultaneous recordings at each of the combinations. The plurality of sensors may comprise first and second sensors and the first and second sensors may be moved alternately to provide the different combinations. The method may comprise normalising the data provided by each of the first and second sensors following a change of position thereof with respect to the data provided by the other of the first and second sensors.

According to a third aspect of the invention, there is provided a method of exploring a region below a surface of the Earth, comprising: using a plurality of sensors at a plurality of locations to obtain seismic data by recording ambient seismic interface waves in a frequency range whose lower limit is greater than or equal to OHz and whose upper limit is less than or equal to substantially 1 Hz; selecting data in at least one time sub-interval of each of at least some of the recordings in accordance with the

quality of the data in the sub-interval; and processing the selected data to obtain a measure of the energy in a frequency band within the frequency range.

The data may be selected in accordance with the energy contained in the sub-interval. The data may be selected from the sub-interval of each recording containing a smallest amount of energy.

According to a fifth aspect of the invention, there is provided a method of exploring a region below a surface of the Earth, comprising: using a plurality of sensors at a plurality of locations to obtain simultaneously recorded seismic data by recording ambient seismic interface waves in a frequency range whose lower limit is greater than or equal to OHz and whose upper limit is less than or equal to substantially 1 Hz; and processing without normalising the data to obtain a measure of the energy in a frequency band within the frequency range.

The method may comprise a combination of at least two of the methods according to the first to fifth aspects of the invention.

The lower limit may be greater than or equal to one of 0.005Hz, 0.001 Hz, 0.01 Hz and 0.1 Hz.

The data may be recorded at each location over a predetermined period. The period may be greater than substantially half an hour.

The method may comprise processing the data to obtain the measure of the energy of at least one of Rayleigh, Love and Scholte waves.

The locations may be spaced apart by at least substantially 100 metres. The locations may be spaced apart by less than substantially five kilometres. The locations may be arranged as a two-dimensional array.

The seismic interface waves may be recorded at or adjacent an interface.

The data may be processed to obtain measures of the energy in a plurality of different frequency bands within the frequency range. The frequency bands may be contiguous or non-overlapping.

The or each frequency band may have a width between 0.001 Hz and 1 Hz.

The frequency band or at least one of the frequency bands may comprise a discrete frequency.

The method may comprise deriving the or each measure of the energy from the amplitudes of the recorded waves in the or the respective frequency band.

The method may comprise deriving the or each measure of the energy from the number of the recorded waves in the or the respective frequency band.

The method may comprise converting the data to the frequency domain. The method may comprise deriving the or each measure of the energy by integrating the converted data over the or the respective frequency band.

The method may comprise normalising the amplitude of the data. The method may comprise converting the amplitude of the data to one of first and second values according to the sign of the amplitude.

The method may comprise filtering the data to attenuate or remove at least some frequencies outside the or each band.

The method may comprise providing a visualisation of the or each measure. The measures may be represented as a map of the region.

According to a second aspect of the invention, there is provided an apparatus arranged to perform a method according to the first aspect of the invention.

According to a third aspect of the invention, there is provided a computer program arranged to control a computer to perform processing in a method according to the first aspect of the invention.

According to a fourth aspect of the invention, there is provided a computer containing or programmed by a program according to the third aspect of the invention.

According to a fifth aspect of the invention, there is provided a computer-readable medium containing a program according to the third aspect of the invention.

According to a sixth aspect of the invention, there is provided transmission across a network of a program according to the third aspect of the invention.

It is thus possible to provide a technique which may be used to explore a subsurface of the earth in a passive or non-intrusive way. This technique is applicable to a very wide range of locations including onshore and offshore. It may even be possible to use this technique in arctic regions. Further, this technique may be used in protected national parks and the like where it is not possible or not permitted to use active seismic sources.

The invention will be further described, by way of example, with reference to the accompanying drawings, in which;

Figure 1 is a diagrammatic sectional view of a region of the earth illustrating the Hymas technique;

Figures 2 and 3 are graphs of amplitude against frequency illustrating the spectra of signals obtained from the Hymas technique of Figure 1 ;

Figure 4 is a cross-sectional diagram of part of the Earth illustrating data acquisition;

Figure 5 is a block diagram of a computer for performing processing of acquired data;

Figure 6 is a flow diagram illustrating a method constituting an embodiment of the invention;

Figure 7 illustrates the effect of a frequency filtering step of the method of Figure 6;

Figure 8 is waveform diagram illustrating the effect of an amplitude normalisation step of the method of Figure 6;

Figure 9 is a diagram showing particle motion for a Rayleigh wave; and

Figure 10 is a graph of depth in wavelengths against relative displacement in arbitrary units for the vertical and horizontal components of Rayleigh waves.

The present technique records and analyses ambient seismic noise. Such noise includes all natural signals (main energy below - 0.5Hz) and all uncontrolled man - made or "anthropogenic" sources (main energy above ~ 0.5Hz). Except for some transient seismic signals (e.g. earthquakes, explosions and so on), these ambient noise records look like incoherent motion of the subsurface. These incoherent signals are waves of different types and surface waves are mainly expected within these recordings due to their characteristic properties.

The present technique mainly records and analyses seismic interface waves (SIW), such as Rayleigh, Love, Scholte waves. In contrast to body waves (longitudinal and shear waves, or in seismology P- and S-waves), which are normally used for hydrocarbon exploration, these SIW propagate along an interface (e.g. the surface or sea bottom). These SIW also have amplitudes and energy below the surface, which are partly higher than their amplitudes directly at the interface. The depth of the highest amplitude depends on the frequency and the propagation velocity of the respective SIW. As a rough rule of thumb, the penetration of SIW increases with decreasing frequency. Active sources usually used for hydrocarbon exploration have their main frequency above 10 Hz (frequency band 1 ). In this frequency range, the penetration of useable amplitudes of SIW will be less than 10 metres (m). Seismologists are using SIW with frequencies far under 0.1 Hz (frequency band 3) to obtain information on the earth's crust and upper mantle. The present technique uses

waves between these two frequency bands (band 2) to derive the information about the subsurface between Om and 8000m depth.

The active excitation of SIW in frequency band 1 is possible with conventional seismic sources (explosion, vibration and so on) or by human noise (e.g. cars, trucks, industry). In band 3, seismologists are using SIW excited by earthquakes. The active excitation of SIW in frequency band 2 is not really possible (except by nuclear explosions). The present technique thus makes use of ambient noise.

In frequency band 2, the SIW are mainly excited by waves and storms on the ocean. It is possible to measure these SIW everywhere in the world (for example as disclosed by Webb 1998: "Broadband seismology and noise under the ocean", in Reviews of Geophysics) and they are usually called "microseisms".

In general, the analysis of amplitudes of recorded seismic signals/waves is very difficult, for example due to the big influence of the individual response function (including the coupling to the earth) of each seismometer. However, these amplitudes contain further or additional information compared with the dispersion of SIW which is normally analyzed by dispersion analysis.

Figure 4 illustrates a first region of the Earth, indicated generally at 10, and a second region of the Earth, indicated generally at 11 , providing examples of regions which may be explored using the present technique. The region 10 is located offshore below the sea bed 12 whereas the region 1 1 is located onshore below the land surface 13. The sea or ocean is shown at 14 with waves 15 propagating along the surface and breaking on a shore 16.

Figure 4 illustrates two techniques for recording seismic interface waves in order to explore the two regions 10 and 1 1. In order to explore the region 10, a plurality of broadband seismometers 20 of sufficient sensitivity is disposed on the sea bed 12 at a plurality of locations, for example arranged as a grid or array above the region 10. The locations of the seismometers 20 may, for example, be chosen to be a two-dimensional array covering the region 10 and with the spacing between adjacent locations being from approximately 100 metres to approximately 5 kilometres. In particular, the

spacing determines the spatial resolution achievable by the present technique as each seismometer provides data relating to the part of the region generally below the location. Typically a square or rectangular grid of locations is chosen. The seismometers may be sensitive to the vertical component of the seismic interface wave or may also be sensitive to other horizontal components.

The seismometers are selected so as to provide sufficient sensitivity to measure the seismic interface waves in the frequency range of interest, which is below substantially 1 Hz. The lower limit of the range may be chosen according to the depth of the region 10 to be investigated. Frequencies down to OHz may be recorded but a lower limit which may well be acceptable in most situations is thought to be 0.005Hz. The seismometers must therefore provide sufficient sensitivity throughout this frequency range.

In order to explore the region 1 1 , onshore seismometers 21 are arranged as a grid in a similar way.

The regions 10 and 1 1 are intended to depict structures illustrated at 22 comprising known hydrocarbon reservoirs or prospects for such reservoirs. The present technique may be used, either alone or in combination with other subsurface exploration techniques, to look for new hydrocarbon reservoirs or to monitor existing reservoirs, for example during production or reservoir emptying.

The seismic interface waves are recorded for a predetermined period at each of the locations of the grid. It is believed that a recording time at each location of a few hours will generally be sufficient and it is expected that most locations will allow a recording time of less than 24 hours. It is believed that a recording time of the order of half an hour will be sufficient in some situations.

Although it is possible for the seismic interface waves recorded by the seismometers to be processed in real time during recording, it is more likely that the recordings will be stored for subsequent processing. The seismic data from the seismometers are therefore typically converted to the digital domain and stored on any suitable storage

medium. The data are subsequently retrieved when convenient and processed so as to analyse the regions 10, 1 1.

The recorded data are processed by an apparatus in the form of a computer, an example of which is illustrated in Figure 5. The computer comprises a central processing unit (CPU) 30 controlled by a program stored in a read-only memory (ROM) 31. A read/write memory in the form of a random access memory (RAM) 32 is provided for temporary storage of data during processing and, if appropriate, for nonvolatile storage of such data, for example including the results of processing. An input interface 33 is provided to allow the data recorded from the seismometers to be entered into the computer and an output interface 34 is provided for supplying the results of the processing. For example, the output interface 34 may provide means for visualising the results of processing.

Figure 6 is a flow diagram illustrating the processing performed by the computer shown in Figure 5. The recorded data are provided at 40, for example by loading the recorded data into the input interface 33. The data are supplied to a filtering step 41 , which provides optional filtering of the data. For example, the step 41 may apply a highpass frequency filter with a suitable cut-off frequency, for example of 0.01 Hz, in order to eliminate or attenuate energy outside the frequency range of interest. Such filtering may be used to eliminate or attenuate strong low frequency amplitude variations originating outside the region 10, 11 of interest. Figure 7 illustrates at 35 the "raw" data recorded by one of the seismometers over approximately a 24 hour period. The curve 36 illustrates the result of filtering.

It is also possible to provide bandpass frequency filtering in the step 41 so as to attenuate or eliminate frequencies outside the frequency range of interest. It is further possible to use a bandpass filter to select a band of frequencies or a discrete frequency of particular interest. The filtering step 41 may provide simultaneous or sequential filtering into different frequency bands or discrete frequencies with the subsequent processing being repeated for each frequency band or discrete frequency, for example so as to provide information about the structure and/or composition at different depths within the region 10, 11 as described hereinafter.

The optionally filtered data are then supplied to a normalising step 42 which, again, is optional and which may be used to provide amplitude normalisation of the recorded data. Such normalisation may improve the data quality, for example to reduce the effects of amplitude variations resulting from different recording times or from specific artefacts, such as earthquakes.

Various normalisation techniques are possible and may be used according to the requirements and according to the quality of the recorded data. For example, in order to remove "strong events" such as earthquakes, amplitude limiting may be applied so as to limit the positive and negative amplitudes to desired thresholds. Any desired degree of limiting may be applied, for example to limit all of the positive and negative peaks in the recorded data to predetermined thresholds. Such normalisation may be taken to the limit in the form of "one bit normalisation". In such normalisation, only the amplitude sign is taken into account so that each peak is limited to either a common positive value or a common negative value according to the sign of the original amplitude. The result of such normalisation is illustrated in Figure 8.

Other, non-limiting, amplitude normalisation may be performed. For example, absolute mean normalisation may be used.

The time-domain data are then supplied to a step 43, where they are converted to the frequency domain so as to provide amplitude-against-frequency spectrum information. Any suitable transformation may be used and a typical example of such a transformation is a fast Fourier transform. The data in the frequency domain are then supplied to a step 44 for calculating the energy in all or part of the frequency range covered by the data. For example, the amplitude value for one or more specific discrete frequencies may be determined in the step 44. Alternatively or additionally, the step 44 may integrate the amplitude information in the data from the step 43 over one or more frequency bands or windows. For example, the energy at a set of discrete frequencies or found by integrating over a set of contiguous or non-overlapping frequency windows may be calculated by the step 44. As described hereinafter, it is believed that the energy at different frequencies or in different frequency bands provides information more directly related to different depths within the region 10, 1 1.

The result of the step 44 is the calculation of the energy at a specific frequency or in a specific frequency band at the location of the seismometer which recorded that part of the data. This may represent the amplitude of the seismic interface wave at a discrete frequency or in a frequency window. In the case where one bit normalisation has been performed in the step 42, the calculation indicates how often a wave of the specific frequency or of a frequency within the specific band occurs within the recorded data at the location of the seismometer. Both of these "measures" therefore provide a measure of the energy in the seismic interface wave at that frequency or in that band at the location of the seismometer.

The calculated measure of energy may be used directly in order to provide information about the region 10, 1 1 below the locations 20, 21. However, the results of the step 44 may alternatively or additionally be supplied to an optional visualising step 45 for providing a visual output, for example in the form of an image on a display and/or a printed hardcopy. This may assist in interpreting the results of the exploration. For example, the values calculated at the same frequency or in the same frequency range for each of the seismometer locations may be displayed as a two-dimensional horizontal map with the individual values being represented at the locations of the seismometers in the horizontal plane.

It is also possible to provide a two-dimensional representation of amplitude plotted against one-dimensional position, for example on the horizontal axis, and frequency or frequency band, for example on the vertical axis. By combining a plurality of such "sectional views" in the orthogonal horizontal spatial direction, a three-dimensional visualisation may be provided. An inversion may be performed to convert frequency to spatial depth so as to provide a two-dimensional or three-dimensional visualisation of the structure or composition of the region 10, 11 below the grid of seismometer locations.

Many visualisation tools exist for providing visual representations of subsurface regions. Any suitable tool may be used in the step 45 in order to provide any desired visualisation based on the energy calculations performed in the step 44.

Any type or types of surface interface waves may be recorded in order to perform the present technique. Examples of such waves are Rayleigh waves, Love waves and Scholte waves.

Figure 9 illustrates the motion of a medium 50 on whose surface 51 a Rayleigh wave is propagating. Cartesian axes are illustrated at X, Y and Z with the wave propagating in the horizontal direction X and the Z axis pointing downwardly into the depth of the medium 50.

The particle motion below the surface when a Rayleigh wave is propagating along the surface 51 is illustrated at 52. Thus, the surface or interface Rayleigh wave is propagated by means of particle motion not just at the surface 51 but also within the medium 50 and each particle repeatedly describes a closed loop.

Figure 10 illustrates depth in wavelengths against relative maximum displacement of particles for a Rayleigh wave and for different values of v. This illustrates that the maximum displacement or amplitude in the vertical direction does not occur at the surface 51 or other interface but instead occurs at a depth of approximately 0.2 wavelengths below the surface.

Although the mechanism underlying the present technique is not fully understood at present, it is believed that the structure and/or composition of the subsurface region of the earth interacts with Rayleigh or other seismic interface waves in a way which is encoded in the amplitude of such waves at the interface. Further, it is believed that the measurable effect may be largely a function of the maximum particle displacement or amplitude of the medium through which the wave is propagating. Although all depths may have an effect on the wave as measured at the surface, the maximum effect for the vertical component of Rayleigh waves, for example, may largely result from the structure or composition of the earth about 0.2 wavelengths below the surface or other interface. Thus, it is believed that the amplitudes of seismic interface waves such as Rayleigh waves measured at the surface and of different frequencies principally provide information about the structure or composition at different depths below the measurement locations.

It is not yet known which mechanism is affecting the seismic interface waves. However, it is currently believed that one or more of the following mechanisms may be responsible for affecting the measured amplitudes:

Scattering of the surface interface waves;

Different attenuations of the surface interface waves;

The effects of different wave velocities;

The effects of different fluids.

Thus, the amplitudes measured at the interface may relate to the structure below the measurement location, to the nature of the material such as whether the material is solid, water or hydrocarbon, or both.

The following describes various examples of the present technique. For the data acquisition, at least two sensors are needed which are sensitive to at least one component. Sensors may be used which are sensitive to three components (for example two horizontal and one vertical). To investigate the Rayleigh wave part of the ambient noise, the vertical component or recording is analysed. To investigate also the Love wave part and the radial part of the Rayleigh wave, it is necessary to know the main direction of the incident surface waves and a simple rotation of the data to this direction may be applied. This direction is mostly related to storm events in the vicinity (a few hundred or thousand kilometres away). Meteorological observations may provide this information. It is also possible to use the free available data of seismological stations to locate these storms or to identify the main direction. If at least three sensors are used for data acquisition, it may be possible to perform an array analysis, which may be of known type, to resolve the main direction of incident waves.

The sensors are placed at several locations within the surveyed area, for example at least two to see differences between these two locations. The locations for measurements within the surveyed area are preferably on an equidistant grid or otherwise evenly distributed. This is not essential and useful results can also be

obtained by irregular survey layouts, which might be chosen e.g. due to topography or prior subsurface information.

There are two possibilities for covering a region to be explored. 1. There are enough sensors available to record with all sensors synchronously so that every measurement location will have one sensor. In this case, no normalization of the data is necessary. This way opens the possibility of combining this method, for example, with ambient noise surface wave tomography (ANSWT) or other array method. 2. There are not enough sensors available so that at least some of the sensors have to be moved to other locations. In this case, 2a a permanent station may be used to normalize the data afterwards. The permanent station may be placed within the surveyed area or in the vicinity of the surveyed area. It is also possible to use publicly available data of seismological observatories in the vicinity (the distance depends on the area but may, for example, be a few 100 km) of the surveyed area. 2b. If there are less sensors available (at least two), it is also possible to work without a permanent station. In this case, for example, a "temporarily permanent" station may be used. For example, with two sensors, data acquisition starts with the sensors at locations A and B. After the first recording time, the second sensor is moved from B to D. Afterwards the first sensor is moved from C to E and so on. By this method, it is possible to cover the entire area. Normalization is described within the processing description hereinafter.

The recording time for each seismometer is typically a few hours or less. This depends also on the desired penetration. Higher frequencies mean less penetration and less recording length. It is preferable for at least part of the recording time to be substantially free from locally generated transient noise.

Another possibility is to record for longer (-days) to measure the full variation of ambient noise amplitudes over time. In this case, the quietest time periods may be selected to compare only the spectra of these time intervals. It is assumed that, after each high noise period, there is a calm period which is comparable in amplitude with the previous one.

After the data acquisition is finished, seismometer records are available of different locations over the surveyed area. For each seismometer record, the following steps are applied. Not all are mandatory.

1. Filtering

2. Calculation of FFT/spectrograms

3. Selection of time periods

4. Normalisation

5. Value calculation

In the following, each step is described in more detail. If there are alternatives in the way of processing the data, this is noted. The following steps can be applied to all three components.

1. Filtering: A frequency filter is applied to focus on a specific frequency band. This could be: a high pass frequency filter with a cut off frequency of around 0.01 Hz to eliminate strong low frequency amplitude variations; a band pass filter for a specific frequency band. If a value for a discrete frequency only is to be calculated (see point 4), than a band pass filter may be applied for this frequency: to obtain values for many frequencies, the processing flow may be repeated with different band pass filters.

2. Fast Fourier Transform FFT to transform the recordings from the time domain into the frequency domain. We can do this for the entire recording (2A) or for shorter times by running windows over the record (2B). In the case of 2B, the window length should be at least a few periods of the investigated wave. This is the calculation of a spectrogram.

3. Selection of time periods (optional but recommend; only in case 2B): The spectrogram allows selection of time periods from the recording. It is possible to select the quietest time period to reduce or eliminate undesirable influences due to transient signals (e.g. earthquakes or other anthropogenic noise). Selection of time period:

There are different methods for selecting data from the calculated spectrogram. Some examples: 1. A time period may be selected where the energy within a defined frequency window is minimal. 2. The amplitudes may be sorted within the spectrogram. This means that, for each frequency, the amplitudes are sorted from low to high. Afterwards, it is possible to select the quietest time periods for each frequency and each time. Finally summing over a time period may be performed to stabilize the data and to obtain only one amplitude spectrum which may be supplied to the next processing step. 3. A further method for the selection of useful time periods can be performed in the time domain before performing the FFT. Thereby the energy may be estimated within every time window (or e.g. calculating the RMS value for such a short time window). Afterwards only the time periods with the lowest energy content are used. 4. Another selection method is described by Groos and Ritter (2008; submitted to Geophysical Journal International) and the Diploma thesis by Joern Groos (2007). The time series is divided into periods of different noise classes.

4. As described hereinbefore, data acquisition may be performed in various different ways to cover the surveyed area with sensors. 1. if for every desired position there is a sensor, it is not necessary to perform a normalization. Instead, the same time periods are selected for each sensor. In the case 2a, the permanent station (own sensor or public sensor) may be used for performing a normalization. The spectra of all recordings may be divided by the time equivalent amplitude spectrum of the recording of the permanent station. Before performing the division, it may be useful to smooth the spectra in advance to give stabilized results. Doing it in this way allows only the difference related to the geology between the different locations to be seen. In case 2b, there is no permanent station for the entire survey, but there are "temporarily permanent" sensors. In this case, it is possible to calculate the difference/quotient between the second sensor and the first (the first sensor is used for the normalisation of the second sensor). After moving the first sensor to the position C for the next recording phase, the second sensor stays at its position and may be used for the normalization as a permanent station. In this way, the change between two sensors is always being estimated. Finally, the change between each sensor and the first station may be calculated so as to provide a normalized map of changes to the first station. The results of an applied normalization are comparable amplitude spectra for each sensor location.

5. Value Calculation: There are some possible methods to calculate a final value for each location:

The simplest way is to use the amplitude value for a specific frequency.

Calculation of an integral of the amplitude spectra over a specific frequency window. This frequency window may be between 0.001 and 1 Hz. The width of the frequency window may be variable and may depend on the data and the desired resolution. The result is one single value for each location.

6. The following processing steps are only to visualize the results:

I. It is possible to put all the calculated amplitude spectra from the step 4 together to obtain a 2D profile or a 3D cube of the subsurface. In this case, the frequency is used as the depth coordinate. To convert the frequency into depth, it is recommend to compare the observed changes in amplitudes with changes in the geological structure (if available) or to do a rough estimate (e.g. depth=c/f ^* 0.25; with c=propagation velocity and f=frequency) or to use every synchronised recording of two different stations and estimate a local dispersion curve (representive of the path between the two sensors) by performing a cross-correlation followed by a frequency time analysis (e.g. Benson et al. 2007). A further inversion of these dispersion curves allows a subsurface model of shear wave velocities to be estimated which may be used for a calculation of eigenfunctions to estimate for each frequency the depth of the highest amplitude (Friedrich & Dalkomo 1995). This information may be used for a very precise depth conversion.

II. The calculated values of each location from the step 5 may be plotted as a map to visualize the results.

It is thus possible to create an image of the rough reservoir structure. Compared to active seismic, this technique is highly cost-efficient and environmentally friendly, but of lower resolution.

In embodiments where "quiet" time periods are selected from the records, it is possible to be more confident that the data are not influenced, or are at least less influenced, by transient signals or other anthropogenic sources. Also, this selection allows data which were not recorded simultaneously to be compared.

Some embodiments allow the horizontal component of Rayleigh waves and Love waves to be analysed. This technique is thus not limited to analysing only the vertical component of Rayleigh waves.

By covering the whole area with sensors, it is also possible to apply other methods of analyzing the data. For example, the same data may be analysed using ambient noise surface wave tomography.

In embodiments which provide local dispersion curves, more precise depth estimation may be provided. It is thus possible to provide improved vertical resolution.

It is possible to produce a rough image of the subsurface with a very inexpenisive method and of relatively good resolution. This technique is independent of the environment, so that it is possible to apply it on mountains, in deep sea, in arctic areas or in protected national parks, where is it is not possible or not allowed to use active seismic sources (especially in this frequency band). The technique may provide additional information and may be used in combination with other methods (e.g. ambient noise tomography).

It is believed that a new exploration technique has been provided and may be used in exploration for hydrocarbon reserves or for related functions, such as monitoring production in known hydrocarbon reservoirs. This technique is relatively inexpensive and it may be possible to make use of already existing passive seismic measurements or recordings.

The technique is of the passive type and may be used, it is believed, anywhere in the world. For example, the technique may be used offshore or onshore or even in arctic regions, for example above ice sheets. It may be possible to provide three-dimensional information about the subsurface structure and/or the technique may provide a direct indication of the presence of hydrocarbons.