WO2014144168A2

WO2014144168A2 - Method and system for seismic inversion

Info

Publication number: WO2014144168A2
Application number: PCT/US2014/028460
Authority: WO
Inventors: Douglas S. SASSEN; Michael E. Glinsky
Original assignee: Ion Geophysical Corporation
Priority date: 2013-03-15
Filing date: 2014-03-14
Publication date: 2014-09-18
Also published as: WO2014144168A3

Abstract

Methods, apparatuses, and systems are disclosed for inverting seismic data. In one example of such a method, a support scale of observed seismic data acquired by a plurality of seismic receivers is estimated. Log data from a well located proximate the plurality of seismic receivers is scaled based on the estimated support scale of the observed seismic data, and the scaled log data is blocked into a plurality of segments in order to obtain a reflectivity model of the well. A source wavelet of the observed seismic data is estimated using the reflectivity model of the well, and a reflectivity model of an area away from the well is determined by performing an inversion on the observed seismic data using the estimated source wavelet.

Description

METHOD AND SYSTEM FOR SEISMIC INVERSION

CROSS REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to and the benefit of United States provisional application number 61/799,275 entitled "Method and System for Seismic Inversion," which was filed on March 15, 2013, and which is hereby incorporated by reference in its entirety for all purposes.

TECHNICAL FIELD

[0002] This disclosure relates generally to geophysical exploration systems, and more particularly to methods and systems of inverting seismic data obtained in geophysical surveys.

BACKGROUND

[0003] Petrochemical products such as oil and gas are ubiquitous in society and can be found in everything from gasoline to children's toys. Because of this, the demand for oil and gas remains high. In order to meet this high demand, it is important to locate oil and gas reserves in the Earth. Scientists and engineers conduct "surveys" utilizing, among other things, seismic and other wave exploration techniques to find oil and gas reservoirs within the Earth. These seismic exploration techniques often include controlling the emission of seismic energy into the Earth with a seismic source of energy (e.g., dynamite, air guns, vibrators, etc.), and monitoring the Earth's response to the seismic source with one or more receivers in order to create an image of the subsurface of the Earth. By observing the reflected seismic wave detected by the receiver(s) during the survey, the geophysical data pertaining to reflected signals may be acquired and these signals may be used to form an image of the Earth near the survey location.

[0004] In addition to creating an image of the subsurface of the earth, the data collected during a seismic survey can be used to generate models of the physical properties of the subsurface, which can in turn be used to make business decisions such as whether to drill a well, and where to do so. SUMMARY

[0005] As described in more detail below, some of the disclosed embodiments provide methods and systems for seismic inversion and/or registration. One example method of inverting seismic data includes the acts of filtering log data from a well located proximate a plurality of seismic receivers with a plurality of scales to generate a plurality of scaled log data sets, blocking each of the plurality of scaled log data sets to generate a plurality of reflectivity models for the log data corresponding to each of the plurality of scales, estimating a plurality of source wavelets, each of the plurality of source wavelets corresponding to a respective one of the plurality of reflectivity models, generating a plurality of synthetic seismic data sets, each respective one of the plurality of synthetic seismic data sets generated from a respective one of the plurality of reflectivity models of the log data and a corresponding one of the plurality of source wavelets, selecting one of the plurality of synthetic seismic data sets based on a tradeoff between matching observed seismic data from the plurality of seismic receivers with the plurality of synthetic seismic data sets and utilizing a small number of reflectivities, determining a regularization parameter by matching reflectivities of the selected synthetic seismic data set with an inversion, calibrated by the regularization parameter, of the observed seismic data from the plurality of seismic receivers proximate the well, and determining a reflectivity model of an area away from the well using the determined regularization parameter and a selected one of the plurality of source wavelets corresponding to the selected synthetic seismic data set as inputs to an inverse model.

[0006] In some embodiments, the regularization parameter may be determined by iterating through a plurality of interim values for the regularization parameter and, for each of the plurality of interim values for the regularization parameter, inverting the observed seismic data from the plurality of seismic receivers proximate the well using respective interim values of the regularization parameter to obtain a plurality of interim reflectivity models of the well, and comparing the plurality of interim reflectivity models of the well with the reflectivities of the selected synthetic seismic data set. The interim value of the regularization parameter used to invert to the interim reflectivity model that most closely matches the reflectivities of the selected synthetic seismic data set may be used to determine the reflectivity model of the area away from the well. The regularization parameter thus determined may dictate the thresholding out of reflectivities that correspond to noise. [0007] In some embodiments, the scale of the selected one of the plurality of synthetic seismic data sets may correspond to an estimated support scale of the observed seismic data, and the regularization parameter may refine a proposed reflectivity model of the area away from the well by applying soft thresholding to eliminate reflectivity coefficients that fall below the estimated support scale of the observed seismic data, and determining the reflectivity model of the area away from the well may further include the act of iterating towards a sparse set of coefficients. In some embodiments, zero crossings of a second time derivative of the natural logarithm of impedances from the log data convolved with a scaled wavelet may be used to block the scaled log data. The plurality of source wavelets may be parsimoniously estimated by constraining each of the plurality of source wavelets to have minimal length and fitting coefficients. Also, the reflectivity model of the area away from the well may be determined using a sparse spike inversion which is parameterized by the regularization parameter.

[0008] In some embodiments, the method may further include the act of determining an uncertainty of the presence and/or amplitude of a reflector in the reflectivity model of the area away from the well, wherein the uncertainty is used as prior information in subsequent Bayesian inversions, as a measure of economic risk, or as a measure of reliability. The uncertainty may be determined by performing a plurality of inversions on the observed seismic data in a Monte Carlo simulation, each seismic inversion including the acts of selecting a realization of an estimated source wavelet using an error distribution function and the source wavelet

corresponding to the selected one of the plurality of synthetic seismic data sets, and selecting a realization of estimated noise level using a noise distribution function and an estimated support scale of the observed seismic data associated with the selected one of the plurality of synthetic seismic data sets.

[0009] In some embodiments, the acts of filtering the log data, blocking each of the plurality of scaled log data sets, and estimating the plurality of source wavelets may be performed for a first set of observed PP seismic data, and may also be independently performed for a second set of observed PS seismic data. The method may also include the act of registering common reflectors in PP and PS seismic data and thereby obtaining estimated Vp/Vs ratios between the common reflectors, and further include the act of predicting a subsurface geophysical parameter based on the registered first and second reflectivity models, and predicting performance of a reservoir based on the predicted geophysical parameter. In some embodiments, determining the reflectivity model of the area away from the well may be done via a joint PP-PS inversion, and, before the joint PP-PS inversion, the second set of observed PS seismic data may be registered to the first set of observed PP seismic data, with this registration including the acts of identifying common primary reflectors in the PP and PS seismic data, identifying potential secondary reflectors in the PP seismic data, registering the PS seismic data to the PP seismic data using the identified common primary reflectors, wherein said registering is constrained by prior knowledge of Vp/Vs ratios, determining a sensitivity of the registration of the PS seismic data to the PP seismic data, updating the registration of the PS seismic data to the PP seismic data based on the determined sensitivity and the prior knowledge of the Vp/Vs ratios, and jointly inverting the PP seismic data together with the registered PS seismic data to obtain a common reflectivity model.

[0010] In some embodiments, the act of determining may be carried out using observed PP seismic data to determine a first, PP reflectivity model of the area away from the well and may also be carried out independently for observed PS seismic data to determine a second, PS reflectivity model of the area away from the well. Further, common reflectors in the first and second reflectivity models may be determined. In some embodiments, the log data may be filtered using one of a Harr wavelet or a Gaussian wavelet.

[0011] Another example method of inverting compressional and converted wave seismic data may include the acts of identifying common primary reflectors in compressional wave (PP) seismic data and converted wave (PS) seismic data, identifying potential secondary reflectors in the PP seismic data, registering the PS seismic data to the PP seismic data using the identified common primary reflectors, wherein said registering is constrained by prior knowledge of Vp/Vs ratios, determining a sensitivity of the registration of the PS seismic data to the PP seismic data, updating the registration of the PS seismic data to the PP seismic data based on the determined sensitivity and the prior knowledge of Vp/Vs ratios, determining an uncertainty associated with the updated registration of the PS seismic data to the PP seismic data, and jointly inverting the PP seismic data together with the registered PS seismic data to obtain a common reflectivity model.

[0012] In some embodiments, the prior knowledge of Vp/Vs ratios may include one or more of known natural ranges of compressional and shear wave speeds, probabilities of said known natural ranges, or Vp/Vs data obtained from log data from a well co-located with seismic sensors that record the PP and PS seismic data. Further, the common primary reflectors may be identified based on log data from a well co-located with seismic sensors that record the PP and PS seismic data.

[0013] In some embodiments, the method may also include determining Vp/Vs ratios between the primary and secondary reflectors based on the updated registration, and determining an uncertainty associated with the determined Vp/V s ratios based on the updated registration. The uncertainty associated with the updated registration may be based on a Bayesian model of the registration and the determined sensitivity, and/or the uncertainty associated with the updated registration may be based at least in part on the prior knowledge of Vp/Vs ratios and the updated registration of the PS seismic data to the PP seismic data.

[0014] In some embodiments, the PP and PS seismic data may be jointly inverted using a multi-stack sparse spike inversion, and the multi-stack sparse spike inversion may have a temporal co-location constraint to promote common reflectors between the PP seismic data and the PS seismic data. A PP-gradient stack, a PP-full stack, and a PS stack may be provided as inputs to the joint inversion. Also, in some embodiments, the registration and/or the updating of the registration may be a multiscale approach.

BRIEF DESCRIPTION OF THE DRAWINGS

[0015] FIG. 1 is a block diagram of a seismic surveying system.

[0016] FIG. 2 is a flow chart that illustrates one embodiment of a method for inverting seismic data.

[0017] FIG. 3 is a flow chart that illustrates one embodiment of a method for filtering and blocking well log data for use in the method shown in FIG. 2.

[0018] FIG. 4 is a flow chart that illustrates one embodiment of a method for selecting a regularization parameter value for use in the method shown in FIG. 2. [0019] FIG. 5 is a flow chart that illustrates one embodiment of a method for determining a reflectivity model for use in the method shown in FIG. 2. [0020] FIG. 6 is a flow chart that illustrates one embodiment of a method for determining an uncertainty of the method shown in FIG. 2.

[0021] FIG. 7 is a flow chart that illustrates one embodiment of a method for registering and jointly inverting PP and PS seismic data for use in the method shown in FIG. 2. [0022] FIG. 8 illustrates an embodiment of a computer system used in a seismic surveying system that is capable of storing and/or processing seismic data, such as to carry out the methods illustrated in FIGS. 2 through 7.

DETAILED DESCRIPTION

[0023] FIG. 1 illustrates one embodiment of a seismic surveying system 100. The seismic surveying system 100 includes one or more seismic sources 102, one or more seismic receivers 103, a data storage 106, and a data processing apparatus 108. The seismic surveying system 100 may be adapted for acquiring seismic data in any of a number of different geological settings. For example, the seismic surveying system 100 may be adapted for seismic acquisition in a land- based or marine-based setting in some embodiments. [0024] The seismic source(s) 102 may be anything that emits seismic energy. For example the sources 102 may include one or more air guns (e.g., for use in a marine towed-streamer acquisition), one or more vibrators (e.g., vibrator trucks for use on land), dynamite, and so forth. In some examples, the seismic sources 102 may be naturally occurring, such as a geological disturbance, background seismic noise, or seismic activity induced by hydraulic fracturing. As illustrated in FIG. 1, in some examples, the seismic sources may provide seismic source data to a data storage device 106. The seismic source data may include, for example, amplitudes, times, positions, and so forth of seismic source activity that can later be correlated with the received seismic traces from the receivers 103.

[0025] Seismic energy emitted by the seismic sources 102 may be detected by one or more seismic receiver(s) 103. Each seismic receiver 103 includes one or more sensors that detect a disturbance of a medium at one or more points in time. For example, a seismic receiver 103 may include a pressure sensor such as a hydrophone in some embodiments. A hydrophone detects amplitudes of a pressure wavefield over time. Another example of a seismic receiver 103 may include a motion sensor, which may detect the motion of particles over time, which, in turn, can be related to the rate of change of a pressure wavefield over time. The particle motion sensor may detect particle motion in one, two, or three directional components. The particle motion sensor may be, for example, a geophone or accelerometer. A motion sensor detects the motion of particles or of an elastic medium over time. A motion sensor may detect velocity,

acceleration, or displacement, or some combination of these, and may do so in one, two, or three directional components. A seismic receiver 103 may alternatively or additionally include other types of sensors.

[0026] The seismic receivers 103 may be positioned proximate the seismic sources 102 during a seismic survey. During the seismic survey, one or more seismic sources 102 may be fired, and the one or more seismic receivers 103 may measure one or more disturbances and may generate one or more traces, which are sequences of measurements over a period of time. In general, each component of each sensor may generate a trace. In some surveys, the seismic receivers 103 may measure different seismic disturbances associated with different wave propagations. For example, the seismic receivers 103 may measure compressional wave seismic data, or shear wave seismic data. The compressional seismic waves detected may result from compressional waves being emitted into the earth and compressional waves being reflected and measured by the receivers, which may be denoted PP seismic data. The same or different receivers may measure shear wave seismic data, which may result from compressional waves being emitted into the earth and shear waves being reflected at interfaces in the subsurface, which may be called converted waves and may be denoted PS seismic data. Each trace may include or may be associated with corresponding positional information, which may be provided by a navigation system (not shown in FIG. 1).

[0027] In some instances, a wellbore may be drilled in an area proximate the seismic receivers 103, in order to directly ascertain information regarding the subsurface in the wellbore. One or more instruments may be lowered into the wellbore in order to measure different properties of the subsurface, such as impedances of various layers, which may be collected into a well log. The impedance values may be calculated from measured density and compressional and/or shear wave velocities measured by the instruments. In some examples, the well log data may be acquired independent of the seismic survey using the sources 102 and receivers 103, whereas in other embodiments, active measurements may be made inside the wellbore during a seismic survey.

[0028] The seismic traces generated by the seismic receivers 103 and the well log information may be provided to the data storage 106 in some embodiments. The data storage 106 may be a local data storage 106 near the seismic receiver(s) 103 and may record seismic traces from a single receiver 103 in some examples, or may be a bulk data storage 106 located at a central station and may record seismic traces from a plurality of different receivers 103 in other examples. The data storage 106 may include one or more tangible mediums for storing the seismic traces, such as hard drives, magnetic tapes, solid state storage, volatile and non-volatile memory, and so forth. In some examples, the seismic traces from the seismic receivers 103 and/or the well log information may bypass the data storage 106 and be provided directly to the data processing apparatus 108 in order to at least partially process the seismic traces in real-time or substantially real-time (e.g., to provide quality control information).

[0029] The data processing apparatus 108 may be any computing apparatus that is adapted to process and manipulate the seismic traces from the seismic receivers 103 and/or the well log information, and, in some embodiments, the seismic source data from the seismic sources 102. The data processing apparatus 108 may be a single computing device, or may be distributed among many computing nodes in some examples. In some examples, different computing apparatuses perform different data processing operations. For example, a first may migrate seismic traces to obtain an image of the earth's subsurface, and a second may perform an inversion on the migrated seismic data to obtain a reflectivity model of the subsurface. An image of interest may be a spatial indication of discontinuities in acoustic impedance or the elastic reflectivity of the subsurface, and may be displayed on a tangible medium, such as a computer monitor or printed on a piece of paper. A reflectivity model of the subsurface may include a series of reflectivities and/or reflectivity coefficients that can be analyzed to determine the composition of different portions of the subsurface. This knowledge may then be used to make decisions about whether, where, when, and how to drill an oil or gas well.

[0030] Turning now to FIG. 2, a flowchart illustrating a method 200 for inverting seismic data is shown. The method 200 may be performed by the data processing apparatus 108 of the seismic surveying system 100 in FIG. 1 based on seismic traces generated by one or more seismic receivers 103 and based on well log information from one or more wellbores.

[0031] In operation 202, a support scale of observed seismic data that was acquired by a plurality of seismic receivers (e.g., seismic receivers 103 in FIG.l) is estimated, as described in more detail below with reference to FIG. 3. In some examples, the observed seismic data may be traces output from the seismic receivers 103, while in other examples, the traces may be migrated, and the support scale may be estimated based on the post-migrated seismic data.

Furthermore, the observed seismic data may include compressional wave (PP) seismic data, converted wave (PS) seismic data, or both PP and PS seismic data. [0032] The support scale of the observed seismic data may include those frequencies in the observed seismic signal (e.g., up to a cutoff frequency) that have a signal to noise ratio greater than one, and may be determined based on the sensitivity of the earth to the seismic source signal.

[0033] In operation 204, log data from a well located proximate the plurality of seismic receivers may be filtered (scaled) based on the estimated support scale of the observed seismic data. For example, the calculated impedances of the log data may filtered by convolving the impedances with a wavelet (e.g., a Gaussian, Harr, or other wavelet) whose characteristics depend on the cutoff frequency of the estimated support scale. In some embodiments, the log data may be preprocessed prior to the filtering - for example a high pass or notch filter may be applied to the log data if the observed seismic data does not have support (e.g., are missing or are noisy) at low frequencies.

[0034] In operation 206, the scaled log data may be blocked into a plurality of segments in order to obtain a reflectivity model of the well. The scaled log data may be blocked by taking the second time derivative of the natural logarithm of the log impedances convolved with a scaled wavelet as per

1 d

2 — W(t, a) * (ln(/mp)) (equation 1) and associating the zero-crossings of that second time derivative with boundaries of respective blocks, where Imp is impedance values calculated form the wellbore, σ is the cutoff frequency of the estimated support scale, and W(t,a) is a wavelet such as the Gaussian or similar wavelet. In some examples, the identified boundaries may be localized by following the zero-crossing points of the second derivative from coarse to fine scales of the well log information. In some examples, the blocking may be done by noise-thresholding wavelet coefficients of a multiscale Haar wavelet transformation of the impedance data and then inverse transforming the

thresholded data back to the original domain.

[0035] Backus averages may be used to fill the blocks, and the reflectivity model of the well may be obtained by calculating impedance contrasts between block boundaries in some examples. Because the blocking may explicitly consider the support scale of the observed seismic data, the well log blocking may be generally analogous to the natural blocking that comes about in a conventional sparse spike type seismic inversion.

[0036] In operation 208, a source wavelet of the observed seismic data may be estimated using the refiectivity model of the well. The source wavelet may be estimated using a wavelet extractor program that compares observed seismic data associated with receivers proximate (e.g., co-located with) the well with a proposed synthetic seismic data that is generated by convolving a proposed source wavelet with the scaled reflectivity model of the well. The wavelet extractor program may minimize differences between the observed seismic and the proposed synthetic seismic, and may in some embodiments provide noise estimates as a byproduct of estimating the source wavelet (e.g., the noise in the observed seismic data may be the difference between the observed seismic data and the extracted source wavelet convolved with the reflectivity model of the well obtained from the blocked log data, and the noise may be caused by, for example, multiples or processing artifacts). In some embodiments, the estimation of the source wavelet in operation 208 may be performed iteratively and in conjunction with estimating the support scale, as described above with reference to operation 202 in order to find the best fit between the observed seismic data and the information obtained from the well log. The tradeoff between the "goodness" of the fit and the complexity of the model, in this example, may be quantified by the Akaike information criterion, from which an optimal tradeoff may be selected. [0037] The well log data may have better resolution and may be less noisy than the observed seismic data and thus, the wavelet extractor program may in some examples take the well log data as accurate or "known" information in order to determine estimates of the noise in the observed seismic data by comparing the observed seismic data with the well log data as if the well log data is substantially free of noise. The wavelet may be estimated parsimoniously by constraining the required estimated source wavelet to have minimal length and fitting

coefficients in some embodiments. The extracted source wavelet and estimated noise levels obtained using the observed seismic data from receivers proximate the well location may in some examples be used to invert seismic data away from the well location, as described in more detail below.

[0038] In operation 210, a regularization parameter may be determined based on the reflectivity of the well and the observed seismic data, as described in more detail below with reference to FIG. 4. The regularization parameter may be a tradeoff term that allows an inversion to substantially fit observed seismic data while not fitting noise. [0039] In operation 212, a reflectivity model of one or more areas away from the well may be determined by performing an inversion on the observed seismic data using the estimated source wavelet and the determined regularization parameter in an inverse model, as described in more detail below with reference to FIG. 5. In some examples, the inversion may operate on a single type of seismic data (e.g., PP seismic data or PS seismic data), whereas in other examples, the inversion may be a joint inversion (e.g., that operates on both PP and PS seismic data

simultaneously to determine a joint PP-PS reflectivity model). In those embodiments where the inversion operates on more than one data set, registration between the datasets may be required in some but not all cases, as described in more detail below with reference to FIG. 7.

[0040] With reference now to FIG. 3, a flowchart illustrating a method 300 for estimating the support scale of observed seismic data (e.g., operation 202 in FIG. 1) is shown. The method 300 may be performed by the data processing apparatus 108 of the seismic surveying system 100 in FIG. 1 based on seismic traces generated by one or more seismic receivers 103 (which may have been migrated) and based on well log information from one or more wellbores. Also, as suggested above, in some examples, the log data may be preprocessed prior to, contemporaneous with, or after the estimation of the support scale. [0041] In operation 302, log data from the well may be filtered and blocked using a first proposed support scale of the observed seismic data - for example a first proposed cutoff frequency. In operation 304, the observed seismic data for a location proximate the well may be compared with a synthetic seismic data set generated by convolved the blocked log data (derived from filtering the log) with the estimated source wavelet, and in operation 306, a goodness of fit from the comparison may be obtained.

[0042] Operations 302, 304, and 306 may each be performed in series for a plurality of different proposed support scales (e.g., iterating from high to low cutoff frequencies, associated with small to large support scales), until no further proposed support scales remain, with each iteration providing an estimate of goodness of fit of the synthetic seismic data sets to the observed seismic. The operations 302, 304, 306 may also iteratively be performed in connection with estimating the source wavelet (e.g., operation 208 in FIG. 2) in some embodiments.

[0043] In some embodiments, a comprehensive scan of support scales and/or source wavelets may be done, whereas in other embodiments, an optimization method may be used to determine the appropriate support scale and/or the appropriate source wavelet. The number of proposed support scales, as well as the frequencies and bandwidths of the proposed support scales may be determined based on estimated noise levels and/or estimated limits of potential bandwidth over noise in the observed seismic data, and the proposed support scales may include large support scales with narrow frequency bandwidth as well as small support scales with high frequency bandwidth.

[0044] In operation 308, the support scale that, when applied to the log data and blocked, gives the best fit (e.g., the most realistic compromise between matching reflectors and the fitting actual data) to the observed seismic data proximate the well location, is selected, and in operation 310 the log data is filtered with the selected support scale and blocked into a plurality of segments to obtain a reflectivity model of the well based on the estimated support scale of the observed seismic data near the well. The resolution of the reflectivity model of the area away from the well (e.g., obtained in operation 212 as described above) may be a function of the support scale selected to scale the log data.

[0045] With reference now to FIG. 4, a flowchart illustrating a method 400 for determining a regularization parameter (e.g., operation 210 in FIG. 1) to use in the inversion of operation 212 is shown. The method 400 may be performed by the data processing apparatus 108 of the seismic surveying system 100 in FIG. 1 based on seismic traces generated by one or more seismic receivers 103 (which may have been migrated) and based on well log information from one or more wellbores. In some embodiments, the regularization parameter may be interpreted as a penalty or trade-off term that, when its value is increased, causes the inversion to produce fewer reflectors in the reflectivity model, and when its value is decreased, causes the inversion to produce more reflectors in the reflectivity model. This regularization parameter may thus allow calibration of the inversion by comparing known data (e.g. the reflectivity model of the well obtained from the blocked log data) with observed data (e.g., the observed seismic data at the well location) for a given location (e.g., at the well), and then taking the relationship between the known and observed data for that location and applying a similar relationship to observed data at other locations. Thus in some embodiments, the regularization parameter may be calculated by comparing the reflectivity model of the well obtained from the blocked log data with a reflectivity model of the well obtained by inverting the observed seismic data associated with a location proximate the well using the estimated source wavelet, and the calculated regularization parameter may be applied to observed seismic data in other locations. The calculated regularization parameter may minimize a difference and/or provide a match between the reflectivity model of the well from the blocked log data and an interim reflectivity model of the well obtained from the observed seismic data. The regularization parameter may thus be calculated by the minimization function

/ = \ W t * R t - S(t) \² + \R \¹ (equation 2) where W(t) is the estimated source wavelet, R(t) is the reflectivity model of the well obtained from the blocked log data, S(t) is the observed seismic data proximate the well location, and λ is the regularization parameter.

[0046] Referring to the method 400 shown in FIG. 4, in operation 402, the observed seismic data for a location proximate the well is inverted using an interim value for the regularization parameter in the inversion. In operation 404, an interim reflectivity model for that location is obtained from the inversion in operation 402. Operations 402 and 404 may be iterated for a plurality of different interim values for the regularization parameter. The interim values for the regularization parameter may be preselected, or may be determined based on the obtained interim reflectivity models (i.e., they may be determined based on dynamic comparisons of the obtained reflectivity models with a reflectivity model of the well from the blocked log information). In operation 406, after the plurality of interim reflectivity models are obtained (or as they are being obtained), the interim reflectivity models may be compared with the reflectivity model of the well that was obtained from filtering and blocking the log data to produce the synthetic seismic data sets. In some embodiments, operation 402 may begin with a very small interim value for the regularization parameter, and, that interim value may be increased for subsequent iterations. As the threshold term is increased, it may increasingly threshold out larger and/or more noise events in the inversion (thereby refining successive interim reflectivity models) until there is a match of the interim reflectivity model and the reflectivity model from the selected synthetic seismic data set, and/or until there is a match of the observed seismic data proximate the well with the interim reflectivity model convolved with the estimated source wavelet. [0047] In operation 408, the value of the regularization parameter that gives the interim reflectivity model that most closely matches the reflectivity model of the well obtained from the blocked log data is selected, and in operation 410, the selected value is used to calibrate the inversion of observed seismic data (e.g., operation 212 in FIG. 2) for one or more areas away from the well location. In general, the regularization parameter may be set ahead of the inversion in operation 212 in FIG. 2, and may be applied for seismic data corresponding to any location in the survey and is not limited to being applied to locations proximate the well. As explained in more detail below, in some embodiments two sets of observed seismic data may be processed according to the methods described herein - such as compressional wave (PP) and shear or converted wave (PS) seismic data. In these embodiments, a regularization parameter value may be calculated for each set of data, or a single regularization parameter may be calculated for both the PP and PS seismic data.

[0048] With reference now to FIG. 5, a flowchart illustrating a method 500 for determining a reflectivity model of an area away from the well using an inverse model (e.g., operation 212 in FIG. 1) is shown. The method 500 may be performed by the data processing apparatus 108 of the seismic surveying system 100 in FIG. 1 based on seismic traces generated by one or more seismic receivers 103 (which may have been migrated) and based on well log information from one or more wellbores.

[0049] In operation 502, a proposed reflectivity model of an area away from the well is generated by performing an inversion on the observed seismic data using the estimated source wavelet and the determined regularization parameter value.

[0050] In operation 504, the proposed reflectivity model is refined by minimizing the norm of the observed seismic data and a proposed synthetic seismic data generated by convolving the proposed reflectivity model of the area away from the well with the estimated source wavelet. The norm may be minimized in some examples using surrogate functionals that orthogonalize the inversion. Furthermore, in some examples, the surrogate functionals may be minimized by an iterative convex optimization method that has global convergence to the norm. In operation 506, which may be performed at substantially the same time as operation 504 (or alternatively may be performed in series with operation 504 in some embodiments), the proposed reflectivity model may be refined by thresholding out reflectivities that correspond to noise and/or fall below the estimated support scale of the observed seismic data. These reflectivities may be thresholded out by applying the regularization parameter determined as per the method 400 illustrated in FIG. 4 and described above; in other words, the amount of thresholding applied in operation 506 may be dictated by the regularization parameter value determined in method 400. As shown in FIG. 5, the operations 504, 506 may be performed one or a plurality of times, and, the result of (iteratively) performing operations 504 and 506 on the initial proposed reflectivity model may be obtaining a final reflectivity model of the area away from the well in operation 508. The final reflectivity model of the area away from the well may in some embodiments be a sparse set of reflectivity coefficients that substantially fit the observed seismic data for that location but that substantially do not fit the noise in the observed seismic data. [0051] The operations 502-508 may be repeated for a plurality of locations away from the well in some embodiments, in order to obtain reflectivity models for a plurality of different locations.

[0052] Turning to FIG. 6, a flowchart illustrating a method 600 for determining an uncertainty of the presence and/or amplitude of one or more reflectors is shown. The method 600 may be performed by the data processing apparatus 108 of the seismic surveying system 100 in FIG. 1 based on seismic traces generated by one or more seismic receivers 103 (which may have been migrated) and based on well log information from one or more wellbores. The operations in the method 600 may be carried out by a Monte Carlo simulation in some embodiments, and the results of the simulation may be one or more uncertainties - for example, an uncertainty of the presence and/or amplitude of one or more reflectors in the reflectivity model of an area away the well obtained from the inversion in operation 212. This uncertainty may be used as prior information in subsequent Bayesian inversions, may be used as a measure of economic risk (e.g., if a well is to be drilled), and/or as a measure of the reliability of the inversion.

[0053] Referring to FIG. 6, the operations 602-606 may be repeated a plurality of times (e.g., 10, 100, 1,000, 10,000, etc.). In operation 602, a realization of the estimated source wavelet is selected using an error distribution function and the estimated support scale of the observed seismic data for one location. In operation 604, a realization of the estimated noise level is selected using a noise distribution function and the estimated support scale of the seismic data for that same location. In some embodiments, the error distribution function and the noise distribution function may be generated as a function of estimated noise the observed seismic data. In operation 606, an inversion is performed on the observed seismic data corresponding to that same location using the source wavelet and noise level realizations selected in operations 602 and 604. In operation 608, the plurality of inversions performed for the plurality of different source wavelet and noise level realizations may be compared to determine the uncertainty of, for example, the presence and/or amplitude of a reflector in a reflectivity model for an area away from the well.

[0054] Referring now to FIGS. 2 through 6, in some embodiments, the methods 200, 300, 400, 500, 600 may be performed for a single type of observed seismic data (e.g., a set of observed PP seismic data, or a set of observed PS seismic data). In other embodiments, however, and as described with reference to FIG. 7, the operations illustrated in FIGS. 2 through 6 may be performed for a plurality of different types of observed seismic data (e.g., observed PP seismic data and observed PS seismic data). For example, the inversion in operation 212 may be a joint PP-PS seismic inversion, which may in some embodiments yield an improved reflectivity model of the subsurface with higher resolution, better accuracy, and less noise than a PP-only or PS- only inversion. Nonetheless, in some embodiments (e.g., where only observed PP seismic data is available), the operations shown in FIGS. 2 through 6 and described herein may be applied to only one type of seismic data to obtain a reflectivity model of the subsurface.

[0055] Turning to FIG. 7, a flowchart illustrating a method 700 for registering and jointly inverting PS and PP seismic data is shown. The method 700 may be performed by the data processing apparatus 108 of the seismic surveying system 100 in FIG. 1 based on seismic traces generated by one or more seismic receivers 103 (which may have been migrated) and based on well log information from one or more wellbores.

[0056] In operation 702, a PP support scale is estimated for a set of observed PP seismic data, and well log data is filtered and blocked with the estimated PP support scale, which may, in some embodiments, be done by following operations 202 to 206 as described above. In operation 704, a PS support scale is estimated for a set of observed PS seismic data, and well log data is filtered and blocked with the estimated PS support scale, which may, in some

embodiments, also be done by following operations 202 to 206 as described above. Operations 702 and 704 may be performed independently of each other, and the PP support scale may be different form the PS support scale due to the different wave propagation of PP and PS seismic waves.

[0057] In operation 706, common primary reflectors (e.g., reflection events that correspond to physical reflectors in the subsurface) between the observed PP and PS seismic data may be identified at the well location. The common primary reflectors may also be determined by considering reflectors in corresponding blocked well log data sets. In some embodiments, the common primary reflectors may also be determined by examining the observed PP and PS seismic data for locations away from the well for matching, easily identifiable locations - for example, the common reflectors in the PP and PS data at the well may be followed along strong continuous events to locations away from the well. The identification of the primary reflectors may be picked automatically, or by a human interpreter in some embodiments.

[0058] In operation 708 potential secondary reflectors are identified in the observed PP seismic data for locations away from the well that may correspond to similar reflectors in the PS seismic data, but that are not easily identifiable in the PS seismic data and/or that do not correspond to easily identifiable reflectors in the data at the well location. [0059] In operation 710 the observed PS seismic data may be initially registered with the observed PP seismic data for locations away from the well based on at least the identified common primary reflectors identified in operation 706. The initial registration may, for example, linearly interpolate the PS data into PP time, and may be constrained by prior knowledge of Vp/V s ratios. The prior knowledge of Vp/V s ratios may include one or more of known natural ranges of compressional and shear wave speeds, probabilities of said known natural ranges, or Vp/V s data obtained from log data from a well co-located with seismic sensors that record the PP and PS seismic data. Following operation 710, in some examples, the observed PP and PS seismic data sets for locations away from the well may be adjusted before proceeding to operation 712 so that the phase and spectra match for the respective data sets. This phase and spectra matching may be similar to techniques used to match phase and spectra in a baseline seismic survey with a monitor seismic survey in time-lapse or 4D seismic exploration.

[0060] In operation 712, the sensitivity (e.g., a likelihood function) of the registration of the observed PS seismic data to the observed PP seismic data for locations away from the well may be determined, and in operation 714, the registration may be updated based on the determined sensitivity and/or the mismatch. Referring to operation 712, the sensitivity may be determined using numerical calculations of derivatives in some examples. The sensitivity may be given by

, . _ P5(t+At)-P5(t)

^ * At

(equation 3) where t is the time of registration points in PS time.

[0061] Referring to operation 714, in some examples, an iterative, non-linear least squares minimization method may be used to update the registration of the observed PS seismic data to the observed PP seismic data based on the determined sensitivity. In some embodiments, and as suggested above, the registration in operation 710 and/or the updates to the registration in operation 714 may be constrained by prior knowledge of Vp/Vs ratios obtained from log data from a well co-located with the seismic sensors that record the PP and PS seismic data.

Constraining the registration and/or the updates to the registration using prior knowledge may help prevent the registration from fitting noise, and may help ensure that the final reflectivity model does not allow impermissible physical behavior in the observed seismic data. [0062] In some embodiments, operations 710, 712 and 714 of the registration process may use a multiscale approach to registering the PP and PS observed seismic data away from the well. This approach convolves the absolute value of the estimated reflectivities with a Gaussian or similar wavelet to upscale the reflectivities. Starting at a large support scale, operations 712 and 714 may be repeated for successively decreasing support scales to minimize the difference (or maximize the match) between the PP and PS reflectivities. Decreasing the scale is equivalent to convolving the reflectivities with a higher bandwidth Gaussian or similar wavelet. The estimated support scale of the observed seismic data may be used to guide the range of scales to use in operations 712 and 714. [0063] In some embodiments, operations 712 and 714 may be performed a plurality of times in order to iteratively update the registration based on successively determined sensitivities. One byproduct of registering the PS seismic data to the PP seismic data may be estimated Vp/Vs ratios between each of the common reflection points. These respective Vp/V s ratios may be used to predict a geophysical parameter of the subsurface, which in turn may be used to predict the performance of a reservoir.

[0064] Once the observed PS seismic data has been (satisfactorily) registered to the observed PP seismic data, the observed PS and PP seismic data may be jointly inverted to obtain a common reflectivity model for the area proximate the well in operation 716. The joint PP-PS inversion may be, for example, a multi-stack sparse spike inversion, described in more detail below. Also, once the observed PP and PS seismic data are registered together for the location proximate the well, a sparse spike type inversion can be used to register common reflectors at areas away from the well, which in turn may allow for joint PP-PS inversions for those locations away from the well.

[0065] In operation 718, an uncertainty associated with the updated registration (e.g., from operation 714) may be determined and/or an uncertainty associated with Vp/Vs ratios that result from the updated registration. The uncertainty may be determined based on prior knowledge of Vp/Vs ratios and/or based on prior knowledge of the uncertainty of the selection of the common reflectors. This prior knowledge may be provided in a Bayesian model with the updating of the registration to obtain an uncertainty of the updated registration and the Vp/V s ratios

corresponding to the updated registration. [0066] As mentioned above, in some embodiments, a multi-stack sparse spike inversion may be jointly performed on observed PP and PS seismic data. The multi-stack sparse spike inversion may be defined by the minimization function fmuiti = h + h + h + \ W₀ {t) * ( ?_! + R₂ + K₃) (t) - (Si + S₂ + 5₃) (t) |² + l₀ |«i + R₂ +

(equation 4) where

/i = |Wi (t) * «_i (t) - 5_i(t) |² + l_i |i|¹ , with i=l,2,3, etc

(equation 5) and W(t) is the estimated source wavelet, R(t) is the reflectivity model of the well obtained from the blocked log data, S(t) is the observed seismic data, λ is the regularization parameter, and f is calculated for each of the input stacks to the multi-stack sparse spike.

[0067] The multi-stack sparse spike inversion may have a temporal co-location constraint that promotes common reflectors between the PP and PS seismic data. The temporal co-location constraint may take advantage of the offset dependence of some sources of error (e.g., multiples) in order to filter out noise events that are temporally misaligned between different offsets. This noise rejection property of the temporal co-location constraint can similarly be used to update the registration of PP to PS seismic data, referring to operation 714 in FIG. 7 and to improve the joint model. In other words, the sensitivity of the multi-stack sparse spike, joint PP-PS inversion to registration between the PP and PS seismic data can be used to update the registration following an initial registration.

[0068] In some embodiments, only one stack of observed PP seismic data and one stack of observed PS seismic data are provided as inputs to the multi-stack sparse spike inversion. In another embodiment, however, a PP-gradient stack, a PP-full stack, and a PS stack are provided as inputs to the joint multi-stack sparse spike inversion, and the PP-gradient, a PP-full, and a PS stacks may correspond to a common gather. These inputs to the multi-stack sparse spike may be linearly independent from one another, and principle component analysis and/or singular value decomposition may be used to orthogonalize and/or create the linearly independent stacks.

[0069] Also, before performing the joint PP-PS multi-stack sparse spike inversion, a "sum stack" may be created by adding together a plurality of linearly independent stacks in order to provide a non-linear correlation model between the linearly independent stacks. The sum stack, together with the linearly independent stacks may together form the inputs for the multi-stack sparse spike inversion.

[0070] In other embodiments of the multi-stack sparse spike inversion, an orthogonalization of the stacks may be done prior to thresholding noise. The thresholding may be done independently on each of the orthogonal stack, and then, following the noise thresholding, the stacks may be projected back into their original basis before updating the reflectivities individually by an optimization method. This process is repeated multiple times until it has converged to a solution.

[0071] FIG. 8 illustrates an embodiment of a computer system 835 capable of processing seismic data, including for example, a system capable of executing the methods 200, 300, 400, 500, 600, 700 illustrated in FIGS. 2 through 7. The computer system 835 illustrated in FIG. 8 may be used as the data processing apparatus 108 in FIG. 1 in some examples.

[0072] In some embodiments, the computer system 835 may be a personal computer and/or a handheld electronic device. In other embodiments, the computer system 835 may be an implementation of enterprise level computers, such as one or more blade -type servers within an enterprise. In still other embodiments, the computer system 835 may be any type of server. The computer system 835 may be onboard a vessel, may be on a remotely controlled drone boat, may be on land in a vehicle, may be in land in a facility, or any other place.

[0073] A keyboard 840 and mouse 841 may be coupled to the computer system 835 via a system bus 848. The keyboard 840 and the mouse 841, in one example, may introduce user input to the computer system 835 and communicate that user input to a processor 843. Other suitable input devices may be used in addition to, or in place of, the mouse 841 and the keyboard 840. An input/output unit 849 (I/O) coupled to the system bus 848 represents such I/O elements as a printer, audio/video (A/V) I/O, etc. [0074] Computer 835 also may include a video memory 844, a main memory 845 and a mass storage 842, all coupled to the system bus 848 along with the keyboard 840, the mouse 841 and the processor 843. The mass storage 842 may include both fixed and removable media, such as magnetic, optical or magnetic optical storage systems and any other available mass storage technology. The bus 848 may contain, for example, address lines for addressing the video memory 844 or the main memory 845.

[0075] The system bus 848 also may include a data bus for transferring data between and among the components, such as the processor 843, the main memory 845, the video memory 844 and the mass storage 842. The video memory 844 may be a dual-ported video random access memory. One port of the video memory 844, in one example, is coupled to a video amplifier

846, which is used to drive one or more monitor(s) 847. The monitor(s) 847 may be any type of monitor suitable for displaying graphic images, such as a cathode ray tube monitor (CRT), flat panel, or liquid crystal display (LCD) monitor or any other suitable data presentation device.

[0076] The computer system includes a processor unit 843, which may be any suitable microprocessor or microcomputer. The computer system 835 also may include a communication interface 850 coupled to the bus 848. The communication interface 850 provides a two-way data communication coupling via a network link. For example, the communication interface 850 may be a satellite link, a local area network (LAN) card, a cable modem, and/or wireless interface. In any such implementation, the communication interface 850 sends and receives electrical, electromagnetic or optical signals that carry digital data representing various types of information.

[0077] Code received by the computer system 835 may be executed by the processor 843 as the code is received, and/or stored in the mass storage 842, or other non-volatile storage for later execution. In this manner, the computer system 835 may obtain program code in a variety of forms. Program code may be embodied in any form of computer program product such as a medium configured to store or transport computer readable code or data, or in which computer readable code or data may be embedded. Examples of computer program products include CD- ROM discs, ROM cards, floppy disks, magnetic tapes, computer hard drives, servers on a network, and solid state memory devices. Regardless of the actual implementation of the computer system 835, the data processing system may execute operations that allow for processing seismic data, including for example the operations illustrated in FIGS. 2 through 7 and otherwise as described herein.

[0078] The apparatuses and associated methods in accordance with the present disclosure have been described with reference to particular embodiments thereof in order to illustrate the principles of operation. The above description is thus by way of illustration and not by way of limitation. Various modifications and alterations to the described embodiments will be apparent to those skilled in the art in view of the teachings herein. Those skilled in the art may, for example, be able to devise numerous systems, arrangements and methods which, although not explicitly shown or described herein, embody the principles described and are thus within the spirit and scope of this disclosure.

[0079] The concepts described herein have broad application. Accordingly, it is intended that all such alterations, variations, and modifications of the disclosed embodiments are within the scope of this disclosure. For example, referring to the method 200 in FIG. 2 and to operations 706-714 in FIG. 7, in some alternate embodiments, a reflectivity model of an area away from the well may be determined independently for observed PP and PS seismic data, and the operations 706-714 may be applied to the independently obtained reflectivity models to register common reflectors therebetween. As another example, the Harr wavelet could be used in some examples during the filtering and blocking of the log data. As still another example, in order to obtain seismic data proximate a well location, the seismic sensors may be co-located at the well location, or, seismic data from sensors at neighboring locations may be propagated towards the well location.

[0080] In methodologies directly or indirectly set forth herein, various steps and operations are described in one possible order of operation, but those skilled in the art will recognize that the steps and operations may be rearranged, replaced, or eliminated without necessarily departing from the spirit and scope of the disclosed embodiments. As mentioned above, for example, in FIG. 2, operation 202 (estimating the support scale) and operation 208 (estimating the source wavelet) may be performed contemporaneously in an iterative fashion in order to find a good fit in some embodiments.

[0081] All relative and directional references (including: upper, lower, upward, downward, upgoing, downgoing, left, right, top, bottom, side, above, below, front, middle, back, vertical, horizontal, and so forth) are given by way of example to aid the reader's understanding of the particular embodiments described herein. They should not be read to be requirements or limitations, particularly as to the position, orientation, or use of the invention unless specifically set forth in the claims. Connection references (e.g., attached, coupled, connected, joined, and the like) are to be construed broadly and may include intermediate members between a connection of elements and relative movement between elements. As such, connection references do not necessarily infer that two elements are directly connected and in fixed relation to each other, unless specifically set forth in the claims.

Claims

CLAIMS What is claimed is:

1. A method of inverting seismic data, comprising:

filtering log data from a well located proximate a plurality of seismic receivers with a plurality of scales to generate a plurality of scaled log data sets;

blocking each of the plurality of scaled log data sets to generate a plurality of reflectivity models for the log data corresponding to each of the plurality of scales;

estimating a plurality of source wavelets, each of the plurality of source wavelets corresponding to a respective one of the plurality of reflectivity models;

generating a plurality of synthetic seismic data sets, each respective one of the plurality of synthetic seismic data sets generated from a respective one of the plurality of reflectivity models of the log data and a corresponding one of the plurality of source wavelets;

selecting one of the plurality of synthetic seismic data sets based on a tradeoff between matching observed seismic data from the plurality of seismic receivers with the plurality of synthetic seismic data sets and utilizing a small number of reflectivities;

determining a regularization parameter by matching reflectivities of the selected synthetic seismic data set with an inversion, calibrated by the regularization parameter, of the observed seismic data from the plurality of seismic receivers proximate the well; and

determining a reflectivity model of an area away from the well using the determined regularization parameter and a selected one of the plurality of source wavelets corresponding to the selected synthetic seismic data set as inputs to an inverse model.

2. The method of claim 1, wherein the regularization parameter is determined by iterating through a plurality of interim values for the regularization parameter and, for each of the plurality of interim values for the regularization parameter, inverting the observed seismic data from the plurality of seismic receivers proximate the well using respective interim values of the regularization parameter to obtain a plurality of interim reflectivity models of the well, and comparing the plurality of interim reflectivity models of the well with the reflectivities of the selected synthetic seismic data set.

3. The method of claim 2, wherein the interim value of the regularization parameter used to invert to the interim reflectivity model that most closely matches the reflectivities of the selected synthetic seismic data set is used to determine the reflectivity model of the area away from the well.

4. The method of claim 2, wherein the regularization parameter dictates the thresholding out of reflectivities that correspond to noise.

5. The method of claim 1, wherein a scale of the selected one of the plurality of synthetic seismic data sets corresponds to an estimated support scale of the observed seismic data.

6. The method of claim 5, wherein the regularization parameter refines a proposed reflectivity model of the area away from the well by applying soft thresholding to eliminate reflectivity coefficients that fall below the estimated support scale of the observed seismic data, and determining the reflectivity model of the area away from the well further comprises iterating towards a sparse set of coefficients.

7. The method of claim 1, wherein zero crossings of a second time derivative of the natural logarithm of impedances from the log data convolved with a scaled wavelet are used to block the scaled log data.

8. The method of claim 1, wherein the plurality of source wavelets are parsimoniously estimated by constraining each of the plurality of source wavelets to have minimal length and fitting coefficients.

9. The method of claim 1, wherein the reflectivity model of the area away from the well is determined using a sparse spike inversion which is parameterized by the regularization parameter.

10. The method of claim 1, further comprising determining an uncertainty of the presence and/or amplitude of a reflector in the reflectivity model of the area away from the well, wherein the uncertainty is used as prior information in subsequent Bayesian inversions, as a measure of economic risk, or as a measure of reliability.

11. The method of claim 10, wherein the uncertainty is determined by performing a plurality of inversions on the observed seismic data in a Monte Carlo simulation, each seismic inversion including the acts of:

selecting a realization of an estimated source wavelet using an error distribution function and the source wavelet corresponding to the selected one of the plurality of synthetic seismic data sets; and

selecting a realization of estimated noise level using a noise distribution function and an estimated support scale of the observed seismic data associated with the selected one of the plurality of synthetic seismic data sets.

12. The method of claim 1, wherein the acts of filtering the log data, blocking each of the plurality of scaled log data sets, and estimating the plurality of source wavelets are performed for a first set of observed PP seismic data, and are independently performed for a second set of observed PS seismic data.

13. The method of claim 12, further comprising registering common reflectors in PP and PS seismic data and thereby obtaining estimated Vp/Vs ratios between the common reflectors.

14. The method of claim 13, further comprising predicting a subsurface geophysical parameter based on the registered first and second reflectivity models, and predicting

performance of a reservoir based on the predicted geophysical parameter.

15. The method of claim 12, wherein determining the reflectivity model of the area away from the well is done via a joint PP-PS inversion, and, before the joint PP-PS inversion, the second set of observed PS seismic data is registered to the first set of observed PP seismic data, said registration comprising:

identifying common primary reflectors in the PP and PS seismic data;

identifying potential secondary reflectors in the PP seismic data; registering the PS seismic data to the PP seismic data using the identified common primary reflectors, wherein said registering is constrained by prior knowledge of Vp/Vs ratios; determining a sensitivity of the registration of the PS seismic data to the PP seismic data; updating the registration of the PS seismic data to the PP seismic data based on the determined sensitivity and the prior knowledge of the Vp/V s ratios; and

jointly inverting the PP seismic data together with the registered PS seismic data to obtain a common reflectivity model.

16. The method of claim 1, wherein the act of determining is performed using observed PP seismic data to determine a first, PP reflectivity model of the area away from the well and is also performed independently for observed PS seismic data to determine a second, PS reflectivity model of the area away from the well, further comprising determining common reflectors in the first and second reflectivity models.

17. The method of claim 1, wherein the log data is filtered using one of a Harr wavelet or a Gaussian wavelet.

18. A method of jointly inverting compressional and converted wave seismic data, comprising:

identifying common primary reflectors in compressional wave (PP) seismic data and converted wave (PS) seismic data;

identifying potential secondary reflectors in the PP seismic data;

registering the PS seismic data to the PP seismic data using the identified common primary reflectors, wherein said registering is constrained by prior knowledge of Vp/Vs ratios; determining a sensitivity of the registration of the PS seismic data to the PP seismic data; updating the registration of the PS seismic data to the PP seismic data based on the determined sensitivity and the prior knowledge of Vp/Vs ratios;

determining an uncertainty associated with the updated registration of the PS seismic data to the PP seismic data; and

19. The method of claim 18, wherein the prior knowledge of Vp/Vs ratios includes one or more of known natural ranges of compressional and shear wave speeds, probabilities of said known natural ranges, or Vp/V s data obtained from log data from a well co-located with seismic sensors that record the PP and PS seismic data.

20. The method of claim 18, wherein the common primary reflectors are further identified based on log data from a well co-located with seismic sensors that record the PP and PS seismic data.

21. The method of claim 18, further comprising determining Vp/V s ratios between the primary and secondary reflectors based on the updated registration, and determining an uncertainty associated with the determined Vp/V s ratios based on the updated registration.

22. The method of claim 18, wherein the uncertainty associated with the updated registration is based on a Bayesian model of the registration and the determined sensitivity.

23. The method of claim 18, wherein the uncertainty associated with the updated registration is based at least in part on the prior knowledge of Vp/Vs ratios and the updated registration of the PS seismic data to the PP seismic data.

24. The method of claim 18, wherein the PP and PS seismic data are jointly inverted using a multi-stack sparse spike inversion.

25. The method of claim 24, the multi-stack sparse spike inversion has a temporal co- location constraint to promote common reflectors between the PP seismic data and the PS seismic data.

26. The method of claim 24, wherein a PP-gradient stack, a PP-full stack, and a PS stack are provided as inputs to the joint inversion.

27. The method of claim 18, wherein the registration and/or the updating of the registration is a multiscale approach.