Processing of seismic reflection data recorded in areas with salt formations [307383]

MASTER THESIS

Processing of seismic reflection data recorded in areas with salt formations

Coordinating teacher Student: [anonimizat].Dr.Ing Ionelia Panea Paraschivoiu Marius Cristian

IUNE 2017

Declaration

I [anonimizat] I now submit for assessment on the programmes of study leading to the award of MSc in Applied Geophysics is entirely my own work and has not been taken from the work of others except to the extent that such work has been cited and acknowledged within the text of my own work. No portion of the work contained in the thesis has been submitted in support of an application for another degree or qualification to this or any institution.

_________________________                                              ___________________

Signed                                                                                    Date

Content

Introduction …………………………………………………………………………………………………………….. 4

Chapter 1

Generation and recording of seismic waves ………………………………………………………………. 5

Types of seismic sources ……………………………………………………………………………………….. 5

Explosive sources ……………………………………………………………………………………………. 5

Surface sources ………………………………………………………………………………………………..7

Types of receivers ………………………………………………………………………………………………… 9

Geophones ……………………………………………………………………………………………………… 9

Receiver arrays ……………………………………………………………………………………………… 10

Chapter 2

Standard processing of seismic reflection data ………………………………………………………….. 14

2.1 Loading the SEG-Y recordings………………………………………………………………………………. 14

2.2 Defining of 2D / 3D geometry………………………………………………………………………………… 16

2.3 Computation and application of static corrections…………………………………………………….. 19 2.4 Frequency filtering ………………………………………………………………………………………………. 20

2.5 [anonimizat] …………………………………………………. 25

2.6 Stacking of traces ………………………………………………………………………………………………… 28

2.7 Time migration ……………………………………………………………………………………………………. 28

Chapter 3

Analysis of seismic reflection data recorded in the Sibiu area ……………………………………. 31

3.1 Description of the study area …………………………………………………………………………………. 31

3.2 Description of the seismic data acquisition ……………………………………………………………… 33

3.3 Description of the seismic data processing ………………………………………………………………. 35

Conclusions …………………………………………………………………………………………………………….. 43

References ………………………………………………………………………………………………………………. 44

Introduction

Seismic reflection surveys are used in studies performed for hydrocarbon exploration. The data are recorded in 2D or 3D surveys and their processing is done following a standard flow (Yilmaz, 2001). Obstacles met in the field along the seismic line are responsible for the irregular geometry, therefore the (x, y)-coordinates for common-depth-points have to be computed in order to obtain the best coverage possible (Panea and Bugheanu, 2016). Depending on the signal-to-noise ratio of the data, the frequency filtering is done using special filters (surface consistent deconvolution, spiking or predictive deconvolution etc). The presence of salt formations into the subsurface affects the imaging of the layers beneath them. Various types of migration can be used to improve the image of seismic section. Jones and Davison (2014) used reverse time migration to improve the seismic illumination of areas in adn around salt bodies. Turning-wave reflections can be used to image vertical and overturned salt flanks (Jones, 2008). The use of depth migration requires velocity models of high accuracy. Lambare et al. (2004) and Lavaud et al. (2004) proposed and used versions of stereotomography to determine velocity models for data recorded in areas with faulted salt formations; these velocity models were used for prestack depth migration.

This thesis contains three chapters. Chapter 1 presents the main types of sources and receivers used for data acquisition. Chapter 2 presents the main processing steps used to obtain a time seismic section. Chapter 3 presents the processing results of a dataset recorded over an area with faulted salt formations. I used two types of migration to improve the image of seismic section in the area with faulted salt layers.

Chapter 1

Generation and recording of seismic waves

Types of seismic sources

A seismic source is a device that releases energy into the ground (Sheriff, 1991). Depending on the position of the source point with respect to the soil surface, there are known explosive sources and surface sources (Milson, 2003).

Explosive sources

Various amounts of dynamite are placed into holes of different depth values. The amount of dynamite and the hole depth depend on the target of the seismic survey. The detonation velocity can be around 6-7 km/s (Telford et al., 1990). Ammonium nitrate and nitrocarbonite (NCN) are the dominant explosive used today. Explosives are packaged in tins or tubes that usually contain 0.5 to 5 kg of explosive. For detonation, electric blasting caps are used; they have a cylindrical form with a length of 4 cm and a diameter of 0.6 cm, containing an embedded resistance and a powder charge mixture that ignites at low temperatures. This caps are placed inside the explosive charges. By passing a large current through the resistance, the generated heat will initiate the explosion and consequently the detonation of the entire charge. The current that causes the blasting cap to explode is derived from a blaster, which is an device that charges a capacitor to a higher voltage using batteries and than discharging the capacitor thru the cap at the desired time (Telford et al., 1990). The holes have to be filled in with dense mud in order to prevent the propagation of seismic energy towards surface (Figure 1.1).

The signal generated by explosive sources have wide frequency content, from 0 to the Nyquist frequency, and minimum phase wavelet. These sources can be used as individual sources or in arrays of sources, depending on subsurface lithology (Figure 1.2). Examples of records obtained using dynamite and Vibroseis are displayed in Figure 1.3. A dynamite record obtained using individual geophones contains all types of waves, significant amount of surface waves and refracted / reflected / diffracted waves.The surface wave body is weaker on the Vibroseis record because the lower frequency of sweep is higher than 8 Hz; the frequencies of the surface waves varies between 2 – 8 Hz or 2 – 12 Hz, depending on the soil surface conditions.

Figure 1.1: Detonation in (left) empty or (right) tamped hole (Cooper and Herrera, 2009).

Figure 1.2: Individual sources or arrays of sources (Cooper and Herrera, 2009).

Figure 1.3: Examples of records obtained using (up) dynamite and (down) Vibroseis (Panea, 2015).

Surface sources

Many allernative energy sources have been developed for use in both land and marine work (Laster, 1985). Surface energy sources are less powerful than explosives and their use on a large scale has been made feasible by vertical stacking methods (Telford et al., 1990). The earliest non-explosive source to gain wide acceptance was the thumper or weight dropper (Telford et al., 1990). Using this method, a weight (around 3t) is dropped from a height of around 3-4 m to generate the energy. In boreholes and in soft marsh where there is the risk of losing equipment, gas guns, air guns and other devices are used. Air guns are sometimes used in bags of water set on the surface of the ground; the coupling with the ground is generally good.

A classification of seismic sources depending on the type of generated energy is given in Enescu and Orban (1979). The most used source in exploration for hydrocarbons is the Vibroseis. A Vibroseis releases all its energy in the shortest amount of time; the signal introduced into the ground is named sweep. A vibrator (usually hydraulic) actuates a steel plate pressed firmly against the ground ,the output wavetrain consists of a sine wave whose frequency increases continuously from 6 to – 50 Hz during the 7 to 21 s sweep (Telford et al.,1990).

Vibroseis sources, like most surface sources, can be used to produce low-energy density, being used in cities and other areas where explosives would cause extensive damage (Telford et al., 1990).

Figure 1.4: Side view of the vibrator system and his major components (Source: www.prakla-seismos.de)

Types of receivers

1.2.1 Geophones

The seismic energy that is arriving at the surface is detected and recorded by geophones. Most common type of geophones being the moving coil one. Figure 1.5 presents the schematic diagram of an moving coil electromagnetic geophone. A cylindrical permanent magnet can be observed in which a circular slot is milled. This slot separates the magnet into two poles (S – central and N – annular). In this slot an dense, consisting of very fine wire coil is inserted that is centered by tree leaf springs (A, B and C). The geophone needs to be placed into the ground in an upright position. Due to the vertical movement of the ground, the magnet will also move, but the coil will tend to stay in place, this movement between the magnetic field and the coil will generate a voltage between the coil’s terminals. The output vollage of the geophone is direct1y proporlional to the strength of the magnetic field of the permanent magnet, the number of lums in the coil,the radius of the coi1 and the velocity of the coil relative to the magnet. Modem high sensitivity geophones have an output of 0.5 lo 0.7 V for a velocity of 1 cm/s of the ground (Telford et al., 1990).

Figure 1.5: Schematic diagram of an moving coil electromagnetic geophone (Telford et al., 1990).

1.2.2 Receiver arrays

Groups of geophones can be laid out in a pattern designed to enhance the useful seismic waves, also called as receiver arrays (Pap,1983). Receiver array responses can be computed from single-sensor recordings in two steps (Panea and Drijkoningen, 2008). First, a number of single-sensor recordimgs equal to the desired number of array elements is summed into one trace then, the summing responses are spatially resampled to the group interval. Array responses from single-sensor data can be obtained after the application of static corrections to the single-sensor data (Figure 1.6). In this way, we improve the frequency filtering, velocity and amplitude analyses results (Panea, 2009). Better surface wave attenuation is obtained using a type of weighted array obtained using the Minimum Variance Distortionless Response (MVDR) beamformer. The single-sensor data are processed to obtain weights that are applied to the single-sensor data before array forming (Figures 1.7 and 1.8).

Any receiver array is defined by two parameters:

– Δxg, array element spacing, is chosen so that the surface waves are not affected by spatial aliasing.

– ΔxG, group interval, is chosen so that the reflected wave is not affected by spatial aliasing.

Figure 1.6: Hard-wired array (left) and standard-array (right) responses for an array with 10 elements spaced at 2.5 m and group interval of 25 m; static corrections applied for a final datum of 0 m and a replacement velocity of 1700 m/s (Panea, 2009).

Figure 1.7: (a) Synthetic seismogram and (b) its (f, k)-amplitude spectrum. 160 single sensors spaced at 2.5 m, time sampling interval is 1 ms (Panea and Drijkoningen, 2008).

.

Figure 1.8: Synthetic seismogram from Figure 1.7 after (a) standard array-forming and (b) MVDR beamforming (Panea and Drijkoningen, 2008).

In industry, various types of receiver arrays are used for seismic data acquisition (Figure 1.9). linear arrays are used in 2D seismic reflection surveys, while non-linear arrays are used in 3D seismic surveys. The best surface wave attenuation is provided by the linear and rectangular receiver arrays, while the circular, cross and fishbone type arrays perform little surface wave attenuation.

Figure 1.9: (a) Linear and (b-e) non-linear geophone arrays.

Chapter 2

Standard processing of seismic reflection data

The results of data acquisition are represented by common-source gathers recorded along linear or crooked profiles, depending on the field conditions, or in 3D seismic surveys. The recorded waves can be minimum or zero phase, depending on the type of seismic source used to generate them. The processing flow and parameters are chosen to process minimum-phase or zero-phase data. The output of data processing is represented by time seismic sections.

Loading the SEG-Y recordings

All common-source gathers recorded in the field are saved in the SEG-Y format. The data saved using this format is composed of two parts, one part is represented by an binary system that contains the seismic traces (an string of amplitude values for each seismic trace) and the second part is represented by an ASCII sistem with the header of the traces were basic informations about this trace are stored (e.g., time sampling interval, number of time samples per trace, X, Y and Z coordinates of shot points and receivers, shot number, trace number, station number, etc). The analysis of the headers is important because they provide important information that is needed for the data processing.

Each common-source gather contains seismic traces that are obtained from the geophones that were active during the generation in one single point on the profile. At this step, we identify the records obtained using dynamite or Vibroseis as seismic sources (Figure 2.1). There are situations in which both types of sources are used in the field and the processing steps have to be chosen depending on the phase of the data, minimum- or zero-phase.

Figure 1.1: Common-source gather obtained using (up) dynamite and (down) Vibroseis.

Defining of 2D / 3D geometry

The input data to this step are represented by the source and receiver (x,y,z)-coordinates. Depending on the type of seismic survey performed in the field, the common-depth-point (CDP) coordinates, fold and offset values are computed using various formulas. The simplest geometry is the one defined for linear profiles (Figure 2.2).

Figure 2.2: Spreadsheets used to define the 2D line geometry

The CDP (x,y)-coordinates are computed by averaging the x-coordinates and y-coordinates for each pair od source-receiver. Offset values represent the distance between a source and a receiver. The fold, M, is computed using:

(2.1)

where, ng is the number of geophones, Δxg is the geophone spacing and Δxs is the source spacing. The fold value varies from 1 to the maximum value, M.

If the data acquisition is performed along crooked profiles, the defining of geometry requires additional information, such as the new position of the processing line and bin size along x- and y-direction (Figure 2.3). Good fold distribution can be obtained using pseudo-3D geometries (Figure 2.4).

The 3D geometry is defined using the source and receiver (x,y,z)-coordinates, inline and xline numbers and all other information that describe a 3D survey. An example of fold and bin distribution is displayed in Figure 2.5.

Figure 2.3: (up) Processing line in crooked-line geometry (Panea and Bugheanu, 2016); (down) the choice of bin size to include all CDPs (white points).

Figure 2.4: Pseudo-3D geometry. Black dots – receivers, white dots – seismic sources (Panea and Bugheanu, 2016).

Figure 2.5: Fold and bin distribution in a 3D geometry (Panea, 2014).

Computation and application of static corrections

Static corrections are calculated and applied to eliminate the time delay caused by topography and low velocity zone (LVZ) from the traveltime of primary reflected waves. For each seismic trace the corrections are represented by the sum of corrections applied to the source and the receiver that generated the analyzed trace. Depending on the position of the reference plane with respect to LVZ, there are used three situations:

the reference plane is above the low velocity zone

In this situation, by applying the static corrections, the R (recever) and S (source) are moved from the soil surface to the reference plane that has a greater elevation than the highest value of the measured dimension along the seismic profile (Figure 2.6). The velocity in the air, v_air, is the velocity of seismic waves that propagate though bed rock, BR. Static corrections, StatC, are expressed in seconds; In this position of the reference plan they have positive values, and they are add to the propagation time of the seismic waves. RC and SC are static corrections for each pair of receiver and source. StatC is the total static corrections applied to a trace.

Figure 2.6: Calculation of static corrections when the reference plane is above the LVZ.

, (2.2)

. (2.3)

the reference plane is below the low velocity zone

In this situation, all static corrections have negative signs, the corrections are substracted from the traveltime of seismic waves to remove the effect of topography and LVZ and, if present, intermediate velocity zone (Figure 2.7).

Figure 2.7: Calculation of static corrections when the reference plane is below the LVZ; IVZ – intermdiate velocity zone, BR – bedrock.

, (2.4)

, (2.5)

. (2.6)

the reference plane intersects the low velocity area

The geophone correction, RC, has the "-" sign as it is substracted from the propagation time of the seismic waves, thereby this correcting moves the geophone from the topographyc surface on the reference plane (Figure 2.8). The source correction, SC , has the "+" sign as it is added to the propagation time of the seismic waves, the source being shifted from the ground surface to the reference plane.

Figure 2.8: Calculation of static corrections when the reference plane intersects the LVZ.

, (2.7)

. (2.8)

Frequency filtering

Frequency filtering is used to remove/attenuate the noisy waves from seismic records. Most used filters are band-pass frequency and fk filters. An example of raw common-source gather is displayed in Figure 2.9. In seismic exploration for hydrocarbons, the surface waves are characterized by low frequencies and apparent velocities. The band-pass frequency filtered version of this record is displayed in Figure 2.10. The use of the fk filters requires the designing of fk polygons on (f,k)-amplitude spectra (Figure 2.11). All unwanted energy is placed outside or inside the fk polygon, depending on the type of the fk filter (accept or reject) used for filtering. Automatic gain control is used to normalize the energy before and after the frequency filtering. An example of filtered record without automatic gain control applied is displayed in Figure 2.12. The zeroed amplitudes at small to zero offsets decrease the accuracy of picked stacking velocities.

Noisy traces have to be zeroed before fk filtering. Static corrections have to be applied before frequency filtering in order to attenuate the aliased energy introduced by topography and lateral velocity variations in the near surface.

Deconvolution can be applied using spiking or predictive versions. Spiking deconvolution is used to increase the frequency content of the filtered data, to increase the seismic vertical resolution. Predictive deconvolution is applied to attenuate/remove the multiples and reverberations. Spiking deconvolution doesn’t attenuate the multiples. Figure 2.13 displays an example of record after band-pass frequency filtering and spiking deconvolution on which primary and multiply reflected waves are seen until traveltimes of 0.8 s. Figure 2.14 displays the same record but after band-pass frequency filtering and predictive deconvolution. By comparing both filtered records, we notice that lots of multiply waves were attenuated after predictive deconvolution.

Figure 2.9: Example of raw common-source gather with clear surface and reflected waves.

Figure 2.10: The band-pass frequency filtered version of record displayed in Figure 2.9.

Figure 2.11: Common-source gather displayed in the (t,x)- and (f,k)-domain (up) before and (down) after fk filtering.

Figure 2.12: Filtered common-source gather without any amplitude correction applied.

Figure 2.13: Filtered common-source gather using band-pass frequency filtering and spiking deconvolution.

Figure 2.14: Common-source gather from Figure 2.13 after band-pass frequency filtering and predictive deconvolution.

Velocity analysis and Normal Move-Out correction

The velocity analysis is used to obtain the 2D or 3D velocity models, depending on the type of seismic data (2D or 3D), used later for stacking of traces, migration and time-to-depth conversion. The input data are represented by filtered CDP-gathers; a CDP-gather is a collection of traces form one CDP point. The interactive velocity analysis is performed at the same time with the Normal Move-Out (NMO) correction. In this way, we verify if the chosen velocity value is correct, flatens the hyperbola of the primary reflected wave or not.

The equation for the NMO correction is obtained using the traveltime equation for a primary reflected wave from a horizontal interface and traveltime at normal incidence (Panea, 2014):

(2.9)

where, V1 is the propagation velocity in the reflective environment above the boundary, x is the position of the geophone which records the reflected wave, and h is the depth of the reflecting boundary. The traveltime at normal incidence is:

(2.10)

By replacing eq. (2.10) in (2.9), we obtain:

(2.11)

By developing in series the above relation and assuming that toVa>> x we obtain:

(2.12)

To calculate the difference between the arrival times of a reflected wave at two geophones R1 and R2 we can write:

(2.13)

(2.14)

(2.15)

(2.16)

(2.17)

If these time differences are calculated for the echo time, to, Δt represents the Normal MoveOut correction.

(2.18) (2.19)

(2.20)

Figure 2.15 displays an example of picking of stacking velocities on a CDP-gather. The accuracy of picked velocities is verified using the NMO correction (Figure 2.16). Flat events on CDP gathers mean corect picked stacing velocities.

Figure 2.15: Velocity analysis before the NMO corection. Red line/white line – stacking velocities, black line – interval velocities.

Figure 2.16: Velocity analysis after the NMO corection. Red line/white line – stacking velocities, black line – interval velocities.

Stacking of traces

In this step, all traces from a filtered and NMOed CDP-gather are stacked into one trace that is saved in a time seismic section. in areas with simple subsurface geology, with horizontal or near horizontal layers, dips smaller than 5-10o, the brute seismic section reflects accurately the subsurface geological structure. Another way to verify the accuracy of stacking velocities is to display both, stacking velocities and time seismic section at the same time. Good velocities are obtained when the stacking velocities follow the structural trends (Figurae 2.17).

Figure 2.17: Stacking velocities (in colours) overlaid on time seismic section.

Time migration

Migration is the process in which a geological boundary is shifted from the measured position in its true position (Ylmaz, 1987). For horizontal-layered medium, an interface has the same position in time or depth before and after migration. In case of dipping limits, the migration has three effects, the migrated limit is moved upward, its dipping angle increases and its length decereases.

All types of migration use velocity information, stacking or interval velocities. Migration can be applied on post-stack data (seismic sections) or pre-stack data (CDP-gathers) using the Kirchhoff or finite-difference methods. The migrated CDP-gathers can be used for a new interactive velocity analysis; the improved velocity model is used to migrate the data, again. The new migrated dataset is stacked and, if necessary, the velocity analysis and migration are repeated until the best results are obtained. An example of NMOed CDP-gather before and after migration is displayed in Figure 2.18. The reflections continuity and amplitude can be improved after pre-stack time migration.

Figure 2.18: NMOed CDP-gather (up) before and (down) after pre-stack time migration.

Post-stack time migration is applied on brute time seismic sections using Kirchhoff or finite-difference methods. Figure 2.19 displays an example of seismic section before and after migration. The reflection continuity is improved after migration.

Figure 2.19: Time seismic section (up) before and (down) after migration using the Kirchhoff method.

Chapter 3

Analysis of seismic reflection data recorded in the Sibiu area

3.1 Description of the study area

The analyzed data were recorded in 2010 by Prospectiuni S.A. in a project perfomed for gas exploration. The seismic line is located in the southern part of the Transylvanian Basin, approximately 20 Km North of the city Sibiu (Figure 3.1), it has a NE-SW orientation (Figure 3.2).

Figure 3.1: Map showing the positions of the the seismic profile (blue). Source map: http://maps.google.com.

The Transylvanian Depression represents a major structural element of the post-tectonic Carpathian Orogen, corresponding to a subsidence and young area that overlies a heterogeneous basement belonging to the branches of the Romanian Carpathians (Paraschiv, 1979). The exploration fo this area started in 1909 when commercial gas deposits were discovered.

Stratigraphically speaking, this depression is developed on a heterogeneous basement consisting of two structural complexes: the folder basement and its post tectonic cover (Sandulescu, Visarion, 1979). We will not discuss the stratigraphic succession belonging to the basement of the Transylvanian Depression, because its relevance for this study.

Figure 3.2: Map showing the positions of the receivers and sources on the seismic profile; blue – receivers, red – sources. Source map: http://maps.google.com.

The sedimentary cover, it its turn, is characteristic for the depresion and may be separated into two sedimentary cycles: Tortonian-Sarmatian and Pliocene.

The Tortonian (1000 – 2000 m) is deposited transgressive, consisting of the following stratigraphic members: the dacite tuff horizont, reaching a few hundred meters; sometimes the tuff begins with conglomerates that can be replaced by sandstone and marly clays; the salt horizont (0-1800 m) having a wide but discontinuous distribution in the basin; the salt appears as pot-sedimentary agglomerations, generating domal and diapir structures; the radiolarian schist horizont, about 10 m thick; the Spiralis marl horizon (Paraschiv, 1979).

The Sarmatian is represented by the Buglovian, Volhynian and Bessarabian. The Buglovian compries the beds developing, as a rule in contuity of sedimentation with the Tortonian, between the Ghiris and the Borsa-Turda-Iclod tuff. The Volhynian-Bessarabian (900 – 1700 m) covers most of the depression area and almost in continuity of sedimentation with Buglovian. The respective deposits are represented by an slternation of marls and sands with sandstone, tuff and limestone intercalation.

The seismic prospecting and the exploration wells carried out in the late ‚70 is indicate several disharmonies in the Middle and Upper Miocene sedimentary series (Paraschiv, 1979).

The Pliocene (150 – 800 m) overlaying unconformably various stratigraphic terms consisting of marls, sands, sandstone, seldom, dolomite and white limestone. Acording to the studies made by Ciupagnea et al. (1970), the Pliocene deposits of the Transylvanian Depression belong to the Pontian and occasionally to the Upper Meotian, conducting to the conclusion that the Transylvanian Depression has been emerged between the end of the Bessarabian and the Upper Meotian (Paraschiv, 1979).

The Quaternary is represented by predominantly fluvial deposits.

3.2 Description of the seismic data acquisition

The analyzed dataset was recorded along a linear profile (Figure 3.2). The seismic energy was generated using explosive sources (dynamite); 1 kg of dynamite/shot point was detonated at 10 m depth. The source spacing was 50 m. The recordings were performed using linear arrays with 12 geophones with the group interval of 25 m. The obstacles found along the line are responsible for the irregular distances between receivers and between sources. The elevation varied along the line. Records characterized by good or poor signal-to-noise ratio are displayed in Figures 3.3 and 3.4. Variations in subsurface geology or incorrect energy generation are responsible for the lack of reflected waves on the record in Figure 3.4. Some seismic waves were recorded with spatial aliasing (Figura 3.5). Aliased energy was introduced by the irregular distance between geophones. This energy appear over the entire wavenumber interval and at frequencies corresponding to reflected waves (up to 60 Hz). Time sampling interval was 2 ms, record length was 5 s.

Figure 3.3: Example of record with clear reflected waves; elevation profile on top.

Figure 3.4: Example of record without clear reflected waves; elevation profile on top.

Figure 3.5: Example of record with spatially aliased arrivals displayed in the (left) time-space and (right) frequency-wavenumber domains.

3.3 Description of the seismic data processing

Data processing was done using a standard flow from industry: loading the common-source gathers in the SEG-Y format, defining the geometry for a linear profile, computation and application of static corrections, frequency filtering, velocity analysis and Normal Moveout correction, stacking of traces and migration.

The input dataset contains a number of 195 records with variable number of traces/record. The time sampling interval was 2 ms and the record length was 5 s. Data acquisition was done using a geophone numbers ranging from 120 to 240. Examples of raw records are displayed in Figures 3.6 and 3.7. The reflected waves are clearer on the spli-spread record (Figure 3.7). The surface waves are characterized by low frequencies and they can be easily attenuated/removed by filtering.

The geometry was defined for a linear profile. The fold is variable along the line due to the acquisition parameters (irregular spacing between receivers and sources). The maximum value is 64 (Figure 3.8). Static corrections were computed using a constant replacement velocity of 1600 m/s.

The surface wave attenuation was done using a band-pass frequency filter applied for 12-60 Hz (Figures 3.9 and 3.10). Figure 3.9 displays the band-pass frequency filtered version of the record from Figure 3.6. Reflected waves are clearer at near offsets after filtering than before it. All head waves seen on the filtered record will be attenuated using a f-k filter. Automatic gain control has to be applied after filtering in order to normalize the amplitudes along the traces. Good surface wave attenuation using the same band-pass frequency filter was obtained for the record displayed in Figure 3.10 (compare Figures 3.7 and 3.10). The reflected waves are clear at zero and near offsets. All head waves will be removed using the f-k filtering. The filtered versions of the records from Figures 6 and 7 are displayed in Figures 11 and 12. Next, the traces from all filtered records were sorted into common-depth-point (CDP) gathers and prepared for the inetractive velocity analysis. The resulted 2D stacking velocity model was used for stacking of traces in order to obtain a stacked section (Figure 3.13). The effect of irregular geometry in the field is observed on the stacked section displayed in Figure 3.14.

Figure 3.6: Raw common-source gather recorded using end-on spread; 120 active geophones/point.

Figure 3.7: Raw common-source gather recorded using split-spread; 240 active geophones/point..

Figure 3.8: Fold variation along the seismic line.

Figure 3.9: Record from Figure 3.6 after band-pass frequency filtering.

Figure 3.10: Record from Figure 3.7 after band-pass frequency filtering.

Figure 3.11: Record from Figure 3.6 after band-pass and f-k filtering.

Figure 3.12: Record from Figure 3.7 after band-pass and f-k filtering.

Two types of migration were used to obtain the final seismic sections. Both use interval velocity models obtained from stacking velocities. Steep dip finite difference migration performs post-stack time migration using f-x spatially-variant finite difference extrapolators. This type of migration provides accurate results up to approximately 70 degrees of dip. It uses vertical and lateraly-variant interval velocity models. The migrated seismic section is displayed in Figure 3.15. The image of seismic section between CDPs 2500-2700 is not well improved after migration. Same image was obtained using the reverse time migration (Figure 3.16). This method performs a post-stack time migration using a reverse-time algorithm in the time-wavenumber domain. It uses vertical and laterally-variant interval velocity models. The migration uses the two-way wave equation and can image dips up to 90 degrees of dip. The use of f-x deconvolution after stacking and migration enhanced some reflections on the time interval 0.4 – 1 s between CDPs 2400-2600 (Figures 3.17 and 3.18)

Figure 3.13: Unmigrated seismic section.

Figure 3.14: Unmigrated seismic section showing the irregular geometry (black flags).

Figure 3.15: Post-stack time-migrated seismic section using finite-difference method.

Figure 3.16: Post-stack time-migrated seismic section using reverse time method.

Figure 3.17: Post-stack time-migrated seismic section using finite-difference method and with f-x deconvolution applied after migration.

Figure 3.18: Post-stack time-migrated seismic section using reverse time method and with f-x deconvolution applied after migration.

Conclusions

I analyzed and processed a seismic dataset recorded over an area with faulted salt formations. The data acquisition was performed along a straight line. The fold variation is due the irregular distances between sources and receivers. Records with poor and good signal-to-noise ratio were recorded in the presence of rough topography. The seismic energy attenuation observed on common-source gathers is due to both, rough topography and complex subsurface geology. Band-pass frequency and f-k filters attenuated very well the coherent noise and the refracted waves. Accurate velocities were obtained on points where clear reflections were observed on the filtered CDP gathers. Two types of migration were used to improve the seismic image. The first one is based on the finite-difference method applied in the f-x domain and the second one, reverse-time migration, is applied in the t-k domain. Both peformed some improvements in the investigated area but clearer reflections are seen after the use of f-x deconvolution on the migrated sections.

References

Cooper, N.M, and Herrera, Y., 2009. 2D – 3D design for land seismic operations including Vibroseis theory and parameter design, Lecture Notes, Mustagh Resources Ltd. and Prospectiuni.

Enescu, D., si Orban, T., 1979. Prospectiuni Seismice I, Universitatea din Bucuresti, p. 297.

Jones, I.F., 2008. A modeling study of pre-processing considerations for reverse-time migration, Geophysics, 73(6), p. T99-T106.

Jones, I.F., and Davison, I., 2014. Seismic imaging in and arounbd salt bodies, Interpretation, 2(4), p. SL1-SL20.

Lambare, G., Alerini, M., and Podvin, P., 2004. Stereotomographic picking in practice, 66th EAGE Conf. And Exhib., Extended Abstract.

Lavaud, B., Baina, R., and Landa, E., 2004. Poststack stereotomography: A robust strategy for velocity model estimation, 66th EAGE Conf. And Exhib., Extended Abstract.

Milson, J., 2003. Field Geophysics, The geological field guide series, Wiley, p. 232.

Panea, I., and Drijkoningen, G., 2008. The spatial data-adaptive minimum-variance distortionless-response beamformer on seismic single-sensor data, Geophysics, 73(5), p. Q29-Q42.

Panea, I. (2009). Array forming in the presence of phase variations, The Leading Edge, 2, p. 216-221.

Panea, I., 2014. Prospectiuni Seismice III, Note de curs, format digital.

Panea, I., and Bugheanu, D., 2016. Analysis of crooked-line 2D seismic reflection data recorded in areas with complex surface and subsurface conditions, Exploration Geophysics, http://dx.doi.org/10.1071/EG15107.

Paraschiv D., 1979. Romanian Oil and Gas Fields, Geological prospecting and exploration, Seris A, No. 13, Bucharest, p. 137- 151.

Pap, A., 1983. Source and receiver arrays, report, Amoco Canada, p. 208.

Sandulescu M., Visarion M., 1979. Consideratii asupra structurii tectonice a fundamentului Depresiunii Transilvaniei. D. S. Inst. Geol. Geof. LXVI/5, Bucuresti.

Sheriff, R.E., 1991. Encyclopedic Dictionary of Exploration Geophysics, SEG, p. 384.

Telford, W.M., Geldart, L.P., and Sheriff, R.E., 1990. Applied geophysics, SEG, p. 744.

Yilmaz, O. 1987. Seismic Data Processing. SEG.

Yilmaz, O., 2001. Seismic data analysis, SEG, p. 2027.