The present invention relates to a method of producing and utilising high resolution impedance logs derived from Vertical Seismic Profiling (VSP) data for use in assessing an oilfield subterranean formation, particularly, but not exclusively, by way of Synthetic

Seismograms.

There are several reasons why it is necessary to analyse and assess the constituents of subterranean formations. For example, when surveying and drilling formations during hydrocarbon exploration or extraction operations it is important to gain a detailed

understanding of the physical characteristics of the formations. This includes information on the relative quantities and nature of any rocks, fluids and gases in the vicinity of the wellbore. This will have a bearing on which recovery techniques and drilling operations will provide an optimum extraction process and allow the prediction of any hazards or other problems within the formation in order to facilitate operations with minimal delays or unexpected complications. To this end, it is important to also establish the relationship between those physical characteristics and their manifestation in terms of geophysical-seismic properties. Establishing those relationships allows modifications of the physical characteristics encountered in the wellbore to be used, to anticipate changes in seismic response due to changes of rock and fluid properties distant from the borehole, in order to evaluate development strategies including new drilling locations.

One way of collecting such formation data is to provide a seismic source (such as an airgun, vibrator, explosives etc.) toward the surface and a number of geophones or other receivers inside the wellbore. In known techniques, this then allows a Vertical Seismic Profile (VSP) of the formation characteristics surrounding the wellbore to be created. This VSP data can then be used in a number of different ways to determine the most appropriate drilling, surveying and exploration operations required.

Synthetic Seismograms are also utilised in order to evaluate, calibrate and interpret characteristics of an oilfield formation within the vicinity, and remote from, the wellbore, in hydrocarbon exploration and extraction operations, for the purposes of quantitatively interpreting surface seismic data in order to plan further development and management of the reservoir. The abovementioned techniques are also applicable in exploration wells located using seismic data which fail to find hydrocarbons, in order to determine reasons for failure.

The abovementioned techniques also apply to any wellbore where surface seismic data is a key element in determining a drilling location.

Known techniques to simulate the surface seismic response for the different rock and fluid scenarios encountered in the wellbore involve using the (compression) P-wave and (shear) S- wave velocity logs and density logs recorded in the well in order to produce Synthetic Seismograms which will provide approximations of the seismic response. However, the recorded velocity and density logs, and hence the Synthetic Seismograms, are subject to errors since the drilling process is inherently injurious to the surrounding fabric of the wellbore and since the velocity and density measurements, upon which the velocity and density logs are based, essentially measure the properties of the damaged rock fabric directly adjacent the borehole wall.

In addition, there are a number of other causes of inaccuracies in the velocity and density measurements taken adjacent to the borehole wall including; fluid invasion, de-stressing, caving, and mud caking.

Another key issue when utilising a very high resolution (relative to seismic) velocity log for surface seismic simulation via Synthetic Seismograms is how to scale the very detailed velocity profile of the log down to the velocity profile "seen" at the seismic scale. Furthermore, the velocity and density logs are recorded against depth. In order for them to be used for seismic purposes they must be converted into velocity and density against time. This process contains its own problems.

One rock property known to be strongly indicative of reservoir quality is the seismic impedance. This is also the property that defines the strength of reflections from a succession of rock layers. Numerically, it is the product of rock velocity and rock density, and so a traditional Synthetic Seismogram computes the succession of reflections (the 'reflectivity series' or just 'reflectivity') corresponding to such a succession of rock layers using this product of velocity and density values obtained from borehole logs. However, known techniques also recognise that the upcoming (reflected) signal in a VSP is a more direct measure of the reflectivity series, being derived from impedance variations rather than velocity and density variations.

The conversion of borehole depth to seismic time is helped by the VSP, which gives a direct measure of the one-way time from the surface source to each of the VSP geophones at known depths in the borehole, and can therefore be used to calibrate the depth-time (or time-depth) conversions. However, this critical calibration hinges on the accuracy of the VSP, which can be compromised by several factors including in particular any dip in the strata penetrated by the borehole, and the loss of bandwidth as the seismic pulse travels through the absorptive layers of the earth.

An aim of the present invention is to allow improved Synthetic Seismograms to be generated, particularly in the reservoir, which are not subject to the same inaccuracies or issues of known techniques described above, and which improve the validity of modifications introduced to represent changes in rock and fluid properties.

A further aim is to derive Synthetic Seismograms directly from VSP data, and to provide means to modify such Synthetic Seismograms to represent changes of rock and fluid properties within and adjacent to a reservoir.

A further aim is to improve the validity of VSP data, particularly where the borehole penetrates dipping strata.

According to the present invention there is provided a method of producing and utilising high resolution impedance logs derived from Vertical Seismic Profile (VSP) data for use in assessing an oilfield subterranean formation, the method comprising: - obtaining Vertical Seismic Profile (VSP) data;

generating a VSP Derived Impedance Log (VDIL) from said VSP data;

inputting VDIL impedance values to an initial rock model to determine or refine the dry rock frame modulus;

modifying impedance values of the VDIL to represent one or more postulated changes in particular properties of the rock or its fluid content in order to create a modified VDIL;

and

generating an improved Synthetic Seismogram from said modified VDIL.

Further features and advantages of the present invention will become apparent from the following description.

Embodiments of the present invention will now be described by way of example only, with reference to the following drawings, in which: -

Figure 1 is a schematic diagram of a known prior art method of producing Synthetic Seismograms;

Figure 2 is a graph illustrating such a Synthetic Seismogram produced by the method of Figure 1 where the Synthetic Seismogram is based on p velocity and density logs;

Figure 3 is a schematic diagram illustrating the methodology of the present invention;

Figure 4 is a schematic illustration of example formations for the purposes of explaining why velocity and density logs contain errors;

Figure 5 is a schematic diagram illustrating the methodology of a further aspect of the present invention where an additional impedance based dip correction step is performed; Figure 6 is a schematic diagram illustrating the methodology of a further aspect of the present invention where an additional reflectivity based dip correction step is performed;

Figure 7 is a schematic flow diagram illustrating an alternative or additional method of extracting S impedance;

Figure 8 illustrates graphs which demonstrate the effect of dip on the transposed display;

Figure 9 illustrates trace graphs which show the results of inverting the transposed data; Figure 10 illustrates a trace graph showing the "corridor" stack obtained by stacking the traces of Figure 9; and

Figure 11 is a schematic diagram illustrating a workflow for providing a measured depth scale VSP impedance log in a deviated wellbore.

Figure 1 shows a prior art methodology of generating Synthetic Seismograms. In particular, it describes a prior art method of inputting data to a rock modeller and generating initial and modified Synthetic Seismograms.

In the prior art methodology of Fig. 1, VSP Data 10 is used to calibrate Velocity Log 20 using the VSP first arrival times which is then combined with the Density Log 22 to create

Impedance Log 24 and a resulting First Synthetic Seismogram 26. The Velocity 20 and Density Logs 22 are also input to an Initial Rock Model Builder generally designated 28. The Initial Rock Model Builder uses measured and empirical data on the in-situ state of the rocks and fluids together with the knowledge of their P and S velocities and density to isolate the value of the Rock Frame Dry Modulus at the wellbore's speciflc location. The process uses the Gassmann equation or similar to relate velocity and density to the rock and fluid properties in order to accomplish this.

The Rock Frame Dry Modulus is then input to a Prediction Block generally designated 30. The Prediction Block combines various combinations of rock and fluid properties as required for any evaluation together with the Rock Frame Dry Modulus. Gassmann or similar equations are then used to calculate new values of P and S velocity and density relating to the known rock property changes. These new values are substituted into the original Velocity and

Density logs and revised acoustic impedance logs are formulated as modified velocity logs 32 and modified density logs 33 which are then used to generate modified impedance logs 34 and modified Synthetic Seismograms 36 in a number of ways including evaluating suitable seismic attributes to characterise such modelled properties.

The outputs in Figure 2 show a typical example of a Synthetic Seismogram produced from p wave velocity and density logs being compared with the local surface seismic data. There are problems with this prior art method for producing Synthetic Seismograms.

Figure 4 illustrates the serious problem of errors introduced into the borehole logs by the drilling process. The logs make their measurement of the material close to the borehole, but this is often not representative of the rock in its undrilled condition. In an effort to counter this problem, the logs are edited by hand in a skilled and labour- intensive process which is occasionally dubious and typically expensive.

Another issue with such known methods is that the velocity and density logs are recorded against depth. In order for them to be used for seismic purposes they must be converted into velocity and density against time. In order to simulate the seismic response, they are typically converted to constant velocity and density equal travel time layers. This process is not straightforward and can result in distortions and aliasing effects.

A further problem with such known methods arises because the borehole velocity log measures at very high frequencies circa 5kHz and greater, whereas the VSP measures are made at seismic frequencies. Typically, Gassmann's equation or something similar is used to compute the rock frame modulus and the revised impedances after changing the fluid properties. This equation is accurate for seismic frequencies but is likely to be compromised at the high frequency measures of the velocity logs input to derive the rock frame modulus.

A further problem with such known methods is clearly evident in Figure 2. This shows the density log (RHOB) 22, the P-wave velocity 20, their product the impedance 24, and the calculated reflectivity. The next stage is to convolve the reflectivity with a plausible pulse shape for the seismic pulse (or 'wavelet'), and thereby to produce the initial Synthetic Seismogram 26 (P-wave Synthetic). Since this pulse shape is not known in the prior art, it is frequently assumed to a Ricker pulse. The result of the convolution is often shown with red and blue half-cycles and these are illustrated under the "P-wave synthetic" heading within Fig. 2. Figure 2 also shows a portion of seismic section across the well. Since the match is clearly not good, the typical response is to vary the pulse shape empirically, with or without some equally arbitrary measure of time-variance to represent the effect of the absorptive earth. This is very unsatisfactory. In particular, it gives very little confidence in the similar operations after the revision of the rock model 30 in corresponding step 36. Yet a further problem with the known methods arise when the procedure is applied to shear (or S) waves. The shear velocity can be particularly difficult to characterize. Being around half that of the p-wave means that conventional logging cannot measure the velocity directly in many sections. Further, although specialist borehole tools create flexural waves in the borehole wall, they are not the same as waves that are created in the seismic process and there are likely to be differences.

As described subsequently, the present invention specifies an alternative method for the production of Impedance logs which mitigates a number of the factors that cause inaccuracies in the Velocity and Density logs.

In Figures 3, 5 and 6 the elements of the workflows which appear above the horizontal line A-A represent prior art process steps.

Figure 3 illustrates the approach of the present invention, which involves obtaining an Impedance log from the VSP, rather than from borehole logs. Specifically, in this method a high resolution Log Scale VSP Derived Impedance Log (VDIL) on a linear time scale is produced from VSP Seismic data.

This log provides an alternative set of impedance values for input to Rock Models and also acts as a comparator to those produced using Velocity and Density logs, improving the assessment of the wider subterranean formation of the oilfield. In addition, the impedance is measured against a seismic linear time scale which greatly facilitates the production of synthetic seismograms. Indeed, a benefit of the invention is that the whole of the process from the measure of the initial seismic response of the in situ rocks and fluids to the estimate of the modified seismic response from the modified rock and fluid content can be conducted in the seismic time domain. This means that there is no requirement to convert from depth to time or the reverse.

While the invention remains reliant on the down-going VSP signal (the first arrivals) for linking the Rock Models to the geology encountered in the borehole, it uses the up-going VSP signal (the VSP reflections) for the calculation of the Impedance log. In one manifestation, this is achieved using a mathematical inversion process, whose object is to convert information on interfaces (such as the strength and character of reflections) into information on layers (such as rocks). Where there are multiple samples of the reflections (as the up-going wave encounters a succession of VSP borehole geophones) these may be stacked in order to reduce borehole noise. A corridor stack is shown at 112 and an inversion step at 114 of Figure 3.

The VSP data may also be used as at 116 to generate reflection stacks which are inverted to impedance in order to predict the depth to reflectors below the bottom of the borehole and for comparison with the alternative velocity and density measures. The corridor stack of VSP reflections can be compared with the synthetic seismogram produced from the velocity and density logs and also with the adjacent surface seismic (as in Figure 2) for the purposes of determining the geological origin of reflection events on the surface seismic, and for embedded wavelet estimation, for example. Where differences are apparent between the inverted VSP reflectivity within the depth range of the borehole and the impedance obtained from the velocity and density logs directly or via synthetic seismograms, and in some approaches, differences between the two may be minimized to optimize the result. In these processes no account is taken of the effect of dip on the VSP reflectivity.

The object of modelling (as at 128 and 130) is to simulate the changes in seismic response that will occur due to changes in fluid content, pressure, change in porosity and other factors, away from the calibration point of the well. Having a measure of the current impedances determined from VSP seismic data means that the "calibration" is provided at the same time and frequency scales as the data to which the results of the modelling are to be applied.

Existing known methods require a downscaling of the log measures to the seismic measure at some point or make the assumption that there is a frequency independent relationship. Gassmann or other equations work at seismic frequencies and inputting a seismic scale impedance value provides a match to "the engine" of the modeller.

However, in order to provide an alternative vehicle for the purposes of modelling the effect of changing rock properties such as fluid content, for example, or to determine the correct impedances in areas of fluid invasion or caving, the high resolution VDIL is of benefit. This log needs to measure the reflections at normal incidence. This is when the incident and reflected seismic 'rays' are coincident and at right angles to the dip of the bed. If this is not so, then corrections need to be applied to the measured reflectivity. The VDIL is a separate, independent measure of impedance which can be compared and contrasted with its velocity and density counterpart. However, the depth of impedance changes as described by the log data may be used as an additional constraint in the process of inverting the VSP reflectivity to VDIL and in converting the time scale VSP impedance log to depth.

Advantageously, the VSP data is used to generate the P & S impedance logs at a resolution sufficient to allow the impedances of the individual layers constituting the reservoir or elsewhere to be characterised. These impedances are used as input into the rock modelling process and to provide modified synthetic seismograms showing the altered seismic response due to changes of fluid type etc. To achieve this, the VSP data must be processed in order to provide the highest possible bandwidth reflectivity and therefore the best possible resolution of Impedance after inversion.

For non-horizontal formations the reflectivity is being measured at non-normal incidence, not at right angles to the dip of the bed, and therefore the VSP data needs to be corrected for Dip. Approaches for achieving this are subsequently outlined with reference to Figures 5 and 6

Figure 5 shows a workflow where corrections for the effect of dip are applied to the VSP impedance log (P and S VDIL 115 in Figure 3), using methods described subsequently, in order to produce the dip corrected impedance log 117.

Figure 6 shows an alternative or supplementary workflow where the correction is applied to the reflectivity 111, using methods described subsequently, in order to produce the dip corrected reflectivity 118 and, through inversion, a dip corrected impedance log (VDIL 117).

Figure 7 shows an alternative approach to obtaining high resolution S VDIL, useful in situations where the P-S-S and P-S modes are not present within the recorded data or are such that they cannot be extracted with sufficient fidelity. The first arrival travel times for the P and direct or mode converted S modes are inverted using depth restraints from the P VDIL and other logs to provide P and S velocity logs. Division of the P VDIL by the P velocity log results in a density log.

In the presence of dip, the prior art process of transposing the VSP data to provide traces that are then inverted can distort the reflections. Figure 8 shows the deconvolved up- going wavefield with a bed dipping at 30 degrees on the right. In the centre is the wavelet and on the right hand side is the transposed display. The wavelet within the transposed data can be seen to be very distorted. The receiver interval is fifteen metres. Interpolation between receiver positions (described subsequently) is one way to minimise the distortion.

In a deviated well the distortion is potentially greater because the reflection points are moving laterally as the receivers move laterally (as well as the effect of dip). Figures 9 and 10 are examples of the process of inversion being applied to transposed and stacked reflectivity traces. Current methods appropriate to surface seismic reflectivity have been applied.

Figure 11 describes the workflow appropriate to providing a normal incident VSP velocity log in a deviated well which is compared with its borehole velocity log counterpart recorded along the axis of the deviated wellbore.

In the context of the present invention the term "Appropriate VSP data" is VSP data obtained in vertical or near vertical wells with the source at the wellhead, VSP data obtained in deviated wells with the source offset from the wellhead, VSP data obtained in vertical or near vertical wells with the source offset from the wellhead, VSP data obtained in deviated wells with the source vertically over the receivers and/or at the wellhead (normal incidence, walk-above, vertical incidence) and subsets of 2D or 3D walkaway data sets where the source is vertically or near vertically above the receiver. In the context of the present invention, the term "impedance" means either P impedance (compression impedance) or S impedance (shear impedance). In the presence of birefringence where shear wave splitting occurs then the shear may be either the fast or slow component or both. Where both P and S impedance logs are derived then the measurements can be combined to provide Vp/Vs and Poisson Ratio logs.

Referring to Figure 3 in more detail, the VSP data 100 is processed to provide a measure of reflectivity 111 which is then inverted to provide a high resolution Log Scale VSP Derived Impedance Log 115 values from which is directly input into the Initial Rock Builder process 128.

Conventional processing steps are generally followed for the pre-processing of the VSP raw data, however data evaluation and parameter selection is key to obtaining high resolution reflectivity. Key steps are in the control of amplitude so that the pre inversion reflectivity

represents only the amount of the p or s energy being reflected from any impedance change relative to the incident p or s energy at that impedance change.

Deterministic deconvolution using data from the incident (down-going) wavefield to design an operator to apply to the coincident up-going wavefield from a reflector at or near that receiver ensures that the amplitude of the resulting wavelet represents only the magnitude of the reflection coefficient at that point. The deconvolution operator should be designed from the shortest possible part of the down-going wavefield consistent with including the incident wavelet. The down-going wavefield may be the p mode for p-p and p-s reflections or the s mode converted from p on transmission higher up the well bore for p-s-s reflections.

As the receiver moves up and away from any reflector then there will be an additional element of transmission from the receiver position to the reflector that is not anticipated by the deconvolution using the down-going wave from that receiver position.

Any changes in the wavelet due to the additional absorption (including peg-leg activity) is likely to be small at depth and is occurring only over the circa 200-300ft of the measurement. If necessary, the additional absorption can be measured between the receiver position and the reflector using prior art methods using the first arrivals and an inverse Q filter applied.

The reflectivity derived from higher receivers is used together with the local surface seismic data for wavelet estimation and other purposes. In this case the Fresnel zone on the VSP can be matched to the Fresnel zone on the surface seismic.

In the presence of dip as discussed above, the reflection point moves up dip away from the well as the receiver recedes from the interface. The prior art process of transposition takes data from one receiver position as the reflection response until the time of the first arrival to the next receiver position is reached. The extraction process then moves to the next trace and does the same again until the next, deeper receiver position and so until the last (deepest) receiver position is reached. From there on down data from the last receiver position is used. In the presence of significant dip this means that there is a time jump as the reflection move from up dip back to the well. In the corridor stack this jump is smeared across the width in time of the chosen corridor. For current uses of such data this effect may be acceptable but for high resolution inversion it is not. Figure 8 demonstrates the effect. The bandwidth of the data is 5-140 hz, the dip is 30 degrees and the receiver spacing is fifteen metres. The wavelet within the transposed data is seen to be distorted. Clearly the effect is greater the wider the receiver spacing. To minimise the problem for this invention the deconvolved up-wave data between adjacent receiver positions is interpolated to provide a constant smaller receiver spacing. In the case of deviated wells with the Normal Incidence geometry or other similar survey then in the presence of significant dip the effect is exacerbated because the source and receiver positions are moving laterally along and above the deviated well bore. Similar interpolation methods as used for vertical well zero offset data are used for deviated well data.

As an alternative approach to interpolation, the deconvolved up-wave data can be VSP/CDP mapped or migrated using prior art methods before selecting the traces for inversion. In the presence of dip these processes place the reflection data in their correct lateral position relative to the borehole prior to inversion. Each "corridor" now represents the reflectivity of the sub surface at further lateral locations which are then inverted prior to selecting the trace or sum of traces that best represent the impedance changes at the borehole.

In one manifestation of the invention the reflectivity as measured by successive receivers higher and higher up the borehole and processed to provide a deconvolved up-going wavefield with the maximum available bandwidth, and after the application of inverse Q filtering, if necessary and appropriate, are inverted to impedance to provide multiple VDILs which are then used individually or stacked to provide the VDIL input to the modeller and to generate subsequent synthetic seismograms. For this approach a longer deconvolution operator may be appropriate up to the two-way travel time from the shallowest receiver used in the inversion to the deepest reflector of interest.

For the inversion step, a Generalised Linear Inversion scheme (GLI) may be employed. The results are illustrated in Figures 9 and 10

As the changing down-going wavelet is measured at each and every receiver position, a time variant wavelet can be employed during the inversion process. Special algorithms are used when combining the low frequency time-depth (velocity) information with the higher frequency (impedance) VSP data. In order to provide the most stable absolute values of impedance, tests are conducted with regard to how much smoothing of the time-depth input is required.

Trace alignment techniques based on successive estimates of the prior art down-going wavefield will improve the stability of the first arrival values.

As an alternative or complementary procedure, multiple reflectivity traces (also referred to as transposed traces) allow a series of VSP derived impedance logs to be produced from zero Fresnel zone (when the receiver is at the reflector) up to the Fresnel zone present on the adjacent migrated surface seismic and beyond. An example of this is shown in Figure 9.

In this example the time scale of the VDIL has been converted to depth using the first arrivals and the data is displayed at high resolution across the reservoir. Applying a semblance based or other technique across the series of such VSP derived impedance logs, particularly in the reservoir zone, allows an optimal subsequent stack of adjacent VSP logs to produce the final log to input into the modelling process or for any other purpose such as comparison with the impedance derived from the velocity and density logs. Such a log is shown in Figure 10 which comprises a sum of all eight of the inverted transposed data of Figure 9.

Alternatively, the impedance trace derived at the receiver depth where the Fresnel zone of the VSP is equivalent to the adjacent surface seismic can be used for, for example, wavelet estimation, and the impedance trace estimated when the receiver is at a particular reflector (therefore having zero Fresnel zone) can be used for fluid substitution modelling.

The high resolution impedance log in the time domain may be converted to the depth domain providing a log on the same depth and scale as the conventional logs to allow comparison with the other depth domain logs such as the sonic log. This is illustrated in Figure 11, a process for achieving this conversion is described later in the methodology.

Therefore, rather than inputting impedances derived by combining separate measures of P-wave and S-wave velocity and density into the Initial Rock Model Builder 128 as in prior art Fig. 1, the P and S impedances derived from VDIL 117 are input into the Initial Rock Model Builder 128. This removes any requirement to input Velocity and Density Logs directly into the Initial Rock Model Builder 128 which in turn avoids having to base the Rock Model Builder calculations on Velocity and Density data which may contain errors or have been edited to remove adverse borehole effects or which may be incomplete.

The Initial Rock Model Builder 128 uses measured and empirical data on the in-situ state of the rocks and fluids together with the knowledge of their P and S impedance to isolate the value of the Rock Frame Dry Modulus at the wellbore's specific location. The process uses e.g. Gassmann equation which relate velocity and density to the rock and fluid properties in order to accomplish this. As only P and S impedance (and no density log information) is input from the VSP logs, the fluid density is determined from other well data. The output data on initial rock frame elastic properties is input to a Prediction Block generally designated 130. The Prediction Block combines various combinations of rock and fluid properties as required for any evaluation together with the Rock Frame Dry Modulus. Gassmann equation or similar is used to calculate new values of P and S impedance related to the known rock property changes. These new values are substituted into the VDIL 134 which are then used to produce new synthetic seismograms 136 in a number of ways including evaluating suitable seismic attributes to characterise such changes.

The VDIL time scale is modified to reflect the changes in velocity predicted by the modelling process. This is done by computing the change in travel time (two way) in any layer that has been altered from top down and thus extend or contract the time scale progressively. The new values of impedance are inserted into the modified time scale VDIL. This provides modified impedance logs in TWT. New reflection

coefficients are computed from the revised log. The new reflection coefficients are then convolved with an appropriate wavelet to produce a new synthetic seismogram showing the effect of the altered rock properties.

In one method in order to provide suitable input to the synthetic seismogram process, the VDIL needs to be on a linear depth scale before inserting the modified impedances.

In order to convert the time based original VDIL to depth a process deployed is to back out the density from the impedance log using the density computes from the VSP data as described earlier. As an alternative, the Gardner velocity-density relationship is used in order to provide a velocity log and then the time-velocity relationship from the integrated log or the VSP first arrivals is used to convert to depth. Such a process is necessary when using the VDIL derived velocity measure to compare with velocity logs recorded in deviated well bores for the purposes of, for example, detection and estimation of VTI anisotropy (see Figure 11).

In another method, the first arrival time depth curve from the VSP data is used followed by correlation of the impedance boundaries on the time to depth converted curve with the same boundaries on other depth scale logs such as the sonic or litholog to provide the final depth scaled VSP log if necessary.

The new Synthetic Seismograms 136 are able to determine how the seismic response will change when the properties of the rocks and fluids change without resorting to the velocity and density logs obtained in the wellbore.

The previously described methodology will provide adequate logs for rock property modelling for a variety of VSP geometry scenarios. Examples include a vertical well with a zero-offset seismic source (including walkaway with one or more source positions close to the wellhead and multiple receivers in the wellbore) or in a deviated well where the seismic sources are vertically above each receiver (including a walkaway source line along the projection of the well bore onto the surface into multiple receivers in the wellbore).

The resolution of the VDIL 134 is higher than that of the local surface seismic data. In this regard, for a typical VSP bandwidth of say around 0-120 Hz, of the deconvolved up-wave at the time-depth curve (the low frequencies coming from the VSP first arrival time-depth measure) then with current inversion schemes layer impedance resolution of three metres or better is to be expected.

The invention therefore extracts seismic reflectivity data from suitable VSP and processes this to provide high resolution impedance information which is sufficient to accomplish the majority of seismically related applications for which borehole velocity and density logs are being used. Dip refers to the situation where certain layers of the formation bed are not entirely level. Where this occurs, the formation bed is said to "dip". If not accounted for in VSP impedance methodologies (which typically assume normal incident measures of reflectivity) this scenario can cause inaccuracies in the information gathered since the up-going seismic wave-field reflected from any dipping bed would be measured at non- normal incidence. In the case of a "walkaway" VSP scenario it may be possible to select an alternative seismic source location from within the data that is positioned above the portion of the dipping formation being analysed in order to ensure that a normal or close to normal incidence reflectivity is achieved at any level. However, this is not possible in a vertical wellbore having a zero-offset source VSP arrangement. In such a zero-offset setup, dip in the formation will "migrate" the point of reflection up dip away from the location of the wellbore as the geophone or other receiver position moves upward away from the formation bed. This in turn shortens the time taken for the seismic wave to pass from the seismic source, be reflected by the formation bed and then be detected at the receiver. This shift in reflection time relative to depth can be used to calculate dip in known ways.

Any significant amount of dip will affect the VSP reflectivity response and therefore the impedance generated from it. The reflectivity must be corrected for the effect of dip in order to maintain accuracy in any resulting impedance logs and hence Synthetic Seismograms. This is achieved by first calculating the amount of dip 113 from the suitably processed VSP data.

The value of the dip at each layer is input into equations such as Shuey or Aki and Richards together with the measured value of the reflectivity at the interface to determine the normal incidence reflectivity. The new value of reflectivity is then inserted into the reflectivity log prior to inversion.

In an alternative approach the measured reflectivity is inverted to impedance and then the values of impedance are modified to account for the effect of dip; VDIL 117. This is done so that changes of impedance across a particular, dipping, boundary produce the required normal incident reflection value. Correcting the inverted data minimises the effect of the wavelet present in the reflectivity and the low frequency portion from the time-depth curve has been included.

As an alternative to, or in addition to, determining the amount for dip from the VSP data, downhole micro electrical resistivity techniques such as Formation Micro Imager FMI logs and dipmeters can be utilised.

Where transmitted mode converted shear or direct shear, when a shear wave source is used on land, is present then the travel times of such arrivals, often from multiple mode converting boundaries, together with the depth of impedance boundaries provided by the p-p VDIL or other borehole logs are jointly inverted to provide interval shear velocities across for example the reservoir area. The same technique can also be used for the p first arrival thus providing a p interval velocity log.

Dividing the p velocities determined as above into the equivalent p impedance intervals will provide a measure of density. The density so obtained can be factored in with the shear velocities to provide shear impedance.

The density log so obtained can be used for the variety of purposes that the prior art density log measured in the borehole wall is. For example, as part of the current synthetic seismogram generation process.

The above described methodology therefore describes an alternative and beneficial way of utilising the VSP Reflectivity Log (or "VDIL"). This has a number of important advantages over prior art methodologies and these include but are not limited to the following:

- There is no requirement for log editing or calibration of either the density or velocity logs in order to provide impedance input for the synthetic seismogram and modelling processes since these logs are not being used. This will save considerable time and effort otherwise needed to prepare such data for this purpose and provides a result with reduced errors. Quality control of the VSP impedance measures is easier and more precise than prior art methodologies for log data.

The method provides a direct measure of impedance, changes of which give rise to the seismic response. The product of two separate log measures, velocity and density, will tend to amplify any errors as both these logs are adversely affected by the same borehole conditions in the same regions. Accuracy is not affected by local borehole conditions such as fluid invasion, de- stressing, mud caking, caving etc.

The elastic properties of the undisturbed rocks and fluids are measured within a Fresnel zone encompassing the well bore. With the receiver adjacent to the reflector and a bandwidth of up to 120 Hz this is typically around three metres, which is beyond any effects due to the drilling process whilst remaining spatially focussed.

The measurements are made at seismic frequencies meaning that scaling of the input data is simplified.

The process from the initial measured VSP reflectivity to the final modified seismogram can be all conducted in the seismic scale time domain.

The log covers the whole of the usable VSP receiver span. Thus VSP derived impedance logs provide a continuous measure of that property from the bottom of the borehole to high up the wellbore where the presence of multiple casing strings may render the data unusable.

Data redundancy allows isolated areas of poor data to be overcome.

The process of obtaining p and s interval velocity logs from the first

arrivals and thus density and s impedance is not generally feasible in deviated wells regardless of how the source is deployed. This is because the receivers are moving laterally along the path of the well and so, unless there is no dip and no lateral variation of the geology, the interval time does not represent a meaningful interval of formation traversed. The above situation also affects the VDIL in a deviated well in that

although the values of impedance, corrected for dip if necessary, are

correct and can be input into the modeller, the time thicknesses of the layers are not meaningful and therefore the VDIL cannot be used to create either an initial synthetic seismogram or altered seismogram after rock and/or fluid modifications in the depth domain. However the original and modified values of impedance from the VDIL together with the known thicknesses from the well data and a knowledge of the deviation can be combined to produce depth scale impedance logs that can then be used to derive conventional synthetic seismograms. Also the impedance can be "mapped" onto a measured depth scale and compared with the borehole velocity log for the purpose of, for example, measuring anisotropy. Although particular embodiments of the invention have been disclosed herein in detail, this has been done by way of example and for the purposes of illustration only. The aforementioned embodiments are not intended to be limiting with respect to the scope of the statements of invention.

It is contemplated by the inventor that various substitutions, alterations, and modifications may be made to the invention without departing from the scope of the invention as defined by the statements of invention.