Land Seismic Acquisition Testing Strategies and Results – Southern Chad, Africa 2013-2015 Cameron Crook, OptiSeis Solutions Ltd. Andrea Crook, OptiSeis Solutions Ltd. Chaminda Sandanayake, Glencore UK Ltd. Paul Stephenson, Gran Tierra Energy Inc. Introduction Over that past four years, Glencore has acquired four 3D seismic surveys covering 1200 km2, 216 2D seismic lines covering 4500 km and twenty-three seismic acquisition tests in Chad (Figure 1). Seismic acquisition testing provided tremendous benefit resulting in improved subsurface imaging, increased operational efficiency and reduced costs. Test results were used to improve operations in real time as data were being acquired, with the cooperation of seismic contractor BGP International (Chad). This paper will provide an overview of the seismic acquisition tests with a specific focus on tests designed to improve imaging in areas with surface laterites/near surface scatters, which impede imaging.

Figure 1: Glencore seismic acquisition programs in Chad – 2011 to 2015

Seismic data acquisition in Chad is conducted using a vibroseis source for logistical reasons. Data quality throughout the area ranges from fair-to-good in the west to excellent in the east. Overall, the quality of data acquired between 2012 and 2015 is superior to previous acquisition programs with better frequency content, improved fault imaging and better reflection continuity at depth. Data quality of previous operator data and the 2012 and 2013 programs was observed to vary from area to area. Glencore geoscientists observed that surface conditions influenced data quality with higher frequency data (70-80 Hz) obtained in areas with silt and sand versus lower frequency data (55-60 Hz) obtained in areas with surface laterites, as mapped on vintage soil maps. In order to further improve imaging in areas with surface laterites/near surface scatters, twenty-three seismic acquisition tests were incorporated into the regular seismic acquisition (Figure 2). Twelve of the tests were acquired during the 2013-2014 seismic acquisition season and were used to optimize acquisition parameters for data acquired during the 2014-2015 seismic acquisition season. Eleven tests were acquired during the 2014-2015 seismic acquisition season for optimization of future seismic programs. These tests include eight vibroseis parameter tests, four maximum offset/record length tests, three sweep effort tests, three 2D Swaths (adding 50 km of high-resolution subsurface data), two noise tests, one receiver array test, one sweep force test and one 3D Cross-Spread test, which provided targeted 3D imaging. Most of the acquisition tests were incorporated into the regular seismic acquisition production with little delay or additional cost (approximately an additional day each field season), so that any potential improvements could be applied immediately.

Figure 2: Seismic acquisition tests acquired by Glencore between 2011 and 2015.

This paper will focus on the tests and results in two areas: Region 1, which has near surface laterites and only fair-to-good data quality; and Region 2, where mixed laterite / laterite free zones resulted in good-to-excellent data quality (Figure 3).

Figure 3: Regional soil map illustrating Region 1 with extensive near surface laterites and Region 2 with limited laterite and good surface conditions. From Audry, Barbery, Boucquier, Bouteyre, Marius, and Vizier et. al.

Soil Maps: 1964 to 1969.

Project Areas Region 1 area is covered in red, iron-rich clay soils with areas of crystalline laterite at the surface (Figure 4). Laterite is a soil and rock type which is rich in iron and aluminium, and is commonly formed in hot and wet tropical areas. Laterites are typically rusty-red due to high iron oxide content. Laterite may form chunks and boulders in the subsurface which may act as scatterers. Additionally, vibrator coupling deteriorates in areas with hard laterite at the surface. Early seismic programs were hampered by unexpected poor data areas. These poor data areas coincided with laterite at the surface or in the near surface. Glencore located and used vintage Chadian soil maps (Audry and Poisot, 1966; Barbery and Bouteyre, 1967; Boucquier and Barbery, 1966; Bouteyre, 1967; Marius and Barbery, 1964 (a) and (b); Vizier and Fromaget, 1969(ab) and (b)) to anticipate problem areas for data quality, especially where surface crystalline laterite was at surface. Where possible, lines were located away from potential bad data areas. Acquisition effort was increased in areas where laterite was unavoidable.

Figure 4: (a) Region 1 soil map (Barbery & Bouteyre, 1967) illustrating extent of surface laterites, (b) sump pit photo illustrating near surface laterite scatters, (c) photo illustrating hard surface laterite.

Overall, the data quality in this area is fair to good with usable frequencies in the 55 to 60 Hz range, and occasionally as high as 80 Hz. Ten seismic lines were acquired in Region 1 during the 2013-2014 seismic acquisition program with four seismic acquisition tests acquired along Line 3. Tests included a vibroseis parameter test, a noise test, a receiver array test, and a sweep effort test. Seven seismic lines were acquired in Region 1 during the 2014-2015 seismic acquisition program with three additional seismic acquisition tests completed on a single line. These tests included vibroseis parameters, sweep force, and sweep effort tests. Data quality in Region 2, which is medium to highly structured, is excellent except under large continuous areas of laterite, where the data quality is only good (Figure 5). Away from laterite, 80 Hz continuous reflectors were seen as deep as 3 seconds with excellent imaging. Under laterite, frequencies and signal-to-noise ratio were significantly reduced, with only 50 Hz along reflectors at 2 seconds. The majority of the seismic lines were acquired in the Region 2, with two 3D seismic surveys and forty-eight 2D seismic lines during the 2013-2014 seismic acquisition program and ninety-four 2D seismic lines during the 2014-2015 seismic acquisition season. Tests in the Region 2 area included a vibroseis parameter tests, a noise test, sweep effort tests, max offset/length tests, 2D swath tests and a 3D cross-spread evaluation. Vibroseis Tests Prior to the start of seismic acquisition recording in these areas, vibroseis parameter tests were conducted in the field in order to select optimal sweep parameters for imaging. Test parameters included the following: Sweep Start/End Frequency, Number of Sweeps, Sweep Length, Force (Drive), Sweep Type (Linear vs. Non-Linear), Start/End Taper, Number of Vibes, Vibroseis Array Length, and Vibroseis Move-up between sweeps. The distribution of vibroseis tests across Region 1 and Region 2 provide excellent reference for designing future 2D and 3D seismic programs. Due to the difficulty imaging through the near surface laterites, for the 2013-2014 seismic acquisition program, a lower frequency 6 – 84 Hz sweep was selected for Region 1; whereas a higher 6 – 96 Hz sweep was selected for the Region 2, which has only limited laterite (Figure 6). In both regions, testing

showed that two sweeps per source station with a move-up between sweeps reduced the airwave. Therefore, move-up between sweeps was incorporated into the 2013-2014 seismic acquisition program.

Figure 5: (a) Region 2 soil map (Marius & Barbery, 1964) illustrating limited extent of surface laterites, (b) photo illustrating good surface ground conditions.

Figure 6: Vibroseis shot records from (a) Region 1, (b) Region 2 illustrating the change in data quality due to near surface laterite in Region 1.

Noise Tests The purpose of the noise tests conducted in Region 1 and Region 2 was to attenuate all crossline noise and record all inline noise unaliased in order to design source and receiver arrays for future seismic programs in areas with laterite. Testing also attempted to identify the presence – if any – of near surface scatterers and the nature of that potential noise. The tests involved acquiring a series of finely sampled (at 4m) 380 m crossline arrays which were sorted and stacked in processing to create finely sampled receiver gathers. Recording of these noise tests was concurrent with production and with minimal additional costs (layout of geophones, etc.). In Region 1, up to 10 km of sources were recorded into the crossline arrays and in Region 2 up to 5 km of sources were recorded into the crossline arrays resulting in a maximum offset of 10 km in Region 1 and 5 km in Region 2. Figure 7 shows a diagram of the 10 crossline receiver arrays with 5 m separation between each array. The 380 m crossline array length, which was selected based on equipment constraints and theoretical modeling, was designed to significantly attenuate crossline noise without attenuating any inline noise (Figure 8) and provides unaliased recording of all noises. Both datasets recorded noises with velocities as slow as 330 m/s at frequencies as low as 8 Hz. Since the recording equipment contained 12 geophone element per group, the 380 m array was recorded as eight separate 12-element arrays and then summed in processing. This method enabled the comparison of different crossline array lengths. Figure 9 illustrates the improvement in imaging achieved with complete crossline noise attenuation and very fine inline spatial sampling.

Figure 7: Noise test array geometry.

In Region 1 noise is predominantly located within an 8-30 Hz bandwidth and in Region 2 noise is predominantly located within an 8-22 Hz bandwidth. Noise velocities in both regions range from 330 m/s to 2150 m/s and are separated into four distinct bands in Region 1 and two distinct bands in Region 2. Ideally, spatial sampling should be fine enough to accurately sample the noise as well as the signal. Based on the noise velocities and frequencies, the ideal spatial sampling interval to record the slowest velocity noise (330 m/s) would be 5.5 m for frequencies up to 30 Hz and 2.75 m for frequencies up to 60 Hz. Given the unreasonably high cost of recording data at this ideal spatial sampling interval (using standard recording equipment), receiver arrays should be used. Although arrays can be designed to reduce or eliminate identified noises, analysis of target wavelengths indicates that arrays should be designed to only attenuate wavenumbers larger than 0.04 (m-1) in order to avoid attenuating signal at shallow targets. Based on the results of the noise test, and the Array Test (below) comparison of final migrated data sets of test line (one with crossline arrays and one with inline arrays), a combination of smaller inline sampling intervals and crossline arrays were recommended and used for the 2014-2015 seismic acquisition program.

- -

Figure 8: Noise test array response for receivers only (a) inline, (b) crossline, and the combined source and receiver (c) inline and (d) crossline array responses. Top: inline horizontal attenuation, Middle: inline vertical attenuation (line represents 30° angle of incidence), Bottom: attenuation filter with black = attenuation at the selected angle, red = average in all directions.

Figure 9: Region 2 noise test illustrating the improvement in data quality achieved by decreasing the spatial sampling interval (a to b or c to d), and increasing the crossline array length (a to c and b to d).

Array Test As observed during the noise tests in Region 1 and Region 2, for 2D acquisition, a crossline array component is beneficial for attenuating crossline noise due to near surface scatters. In order to test inline and crossline receiver array designs, an Array Test was conducted in Region 1 on the same line as the noise test (Line 003). This test involved laying out alternating inline and aerial receiver arrays along the entire length of the line. The linear receiver array was 12 geophones over 25 m and the aerial receiver array was 12 geophones on a diagonal (25 m inline x 45 m crossline) (Figure 10). The purpose of this test was to examine the feasibility of laying out crossline receiver arrays and compare the data quality achieve with a linear array versus an aerial array. It was expected that the aerial array would improve imaging by reducing crossline noise (Figure 11).

Figure 10: Array test geometry with alternating linear and aerial receiver arrays.

The array test was acquired as part of the regular production data, which reduced the costs of this acquisition test and processing. The data were sorted in processing into two datasets: one with only linear receiver arrays and one with only aerial receiver arrays (Figure 12). Final processed migrated data for both inline and crossline arrays were compared at the same location, and indicated that aerial arrays provide better crossline noise attenuation in areas such as Region 1, where significant near surface laterites are present. Signal-to-Noise is also improved in the 10-25 Hz bandwidth, however, at frequencies above 60 Hz, linear receiver arrays resulted in better imaging of shallow horizons and some improvement along steeply dipping reflectors. Based on these results it was recommended that aerial arrays be used in Region 1, but linear arrays be used in areas such as Region 2, where frequencies in excess of 80 Hz are more common. This method was successfully implemented during the 2014-2015 seismic acquisition program with improved imaging obtained with the aerial receiver arrays in areas with laterite.

Figure 11: Combined source and receiver array responses for (a) a linear inline receiver array (a) and (b) an aerial receiver array. Top: inline horizontal attenuation, Middle: inline vertical attenuation (line represents 30° angle of incidence), Bottom: attenuation filter with black = attenuation at the selected angle, red = average in all directions.

Figure 12: Shot record comparing a linear receiver array (a) and an aerial receiver array (b). Note the attenuation of aliased noise with aerial receiver array.

Sweep Effort Tests For the first two programs acquired in Chad in 2012 and 2013, five vibrators were used as a source, originally as an attempt to improve low signal-to-noise observed on the company’s first seismic records in Chad. Subsequent shooting in the following seasons used four vibrators, as data quality was found to be comparable with the one less vibrator and costs were reduced significantly. The 2013-2014 seismic acquisition program was acquired with 25m receiver station intervals and 50m source station intervals with two vibroseis sweeps per source station. These parameters were similar to the parameters used by previous operators. Increasing the fold and acquiring the data with symmetric sampling (Vermeer, 2012) by reducing the source interval to 25m was postulated to improve data quality. However, with two sweeps per 25m source station, the number of sweeps on the program would have doubled, increasing the cost of the acquisition program. In theory, decreasing the number of sweeps will decrease the strength of the signal (Meunier, 2011); however, reducing the source station interval from 50 m to 25 m will double the fold, so the signal strength difference between one versus two sweeps (or two vs. four) should be negligible. Based on this information sweep effort tests were proposed to determine if imaging improvements could be obtained by acquiring data with a smaller source interval (higher fold) while optimizing costs by reducing the sweep effort at each source station. Two data areas were selected for testing: one in Region 1 and one in an area of excellent data quality within Region 2. The tests involved acquiring a line of 2D data with two un-summed sweeps every 25 m and then separating the data in processing into: (1) one sweep every 25 m and (2) two sweeps every 50 m. Test results in both areas showed improved imaging with a single sweep every 25 m vs. two sweeps every 50 m. A comparison of the raw shot gathers for one versus two sweeps shows that random noise attenuation is better on the two sweep record versus the one sweep record. However, in good data areas, the slight increase in random noise due to a single sweep is countered by the large improvement obtained in imaging steep dips and high frequencies when a smaller, 25 m, source station interval is used (Figure 13). Based on the test results, a single sweep, 25 m source station interval was recommended for use in good data areas and a two sweep, 25 m source station interval was recommended for use in difficult data areas. During the 2014-2015 seismic acquisition season, single sweep data was acquired in Region 2 with excellent results. In Region 1 two sweeps every 30 m were used; however, during the 2014-2015 seismic acquisition program, an additional sweep effort test was conducted to compare two sweeps every 30 m versus a single sweep every 15 m. In addition, one of the 3D surveys was acquired with two unsummed sweeps in order to test the effect of one vs. two sweeps on 3D data.

Figure 13: Comparison of two adjacent lines in Region 2: (a) single sweep with a 25 m source station interval, (b) two sweeps with a 50 m source station interval.

Benefits of incorporating seismic acquisition tests into regular production As a result of the successful test programs during the 2013-2014 seismic acquisition program, eleven additional seismic acquisition tests were acquired during the 2014-2015 seismic acquisition program with excellent results. Some of the tests, such as the 2D swath tests and 3D cross-spread test were acquired at minimal cost and identified new methods of expanding the use of seismic in Southern Chad. Since the swath and cross-spread lines were already being acquired as regular 2D lines, the only additional costs were those related to maintaining a larger active spread and the costs for processing the additional swath and cross-spread data. In areas where parallel 2D lines were planned, concurrent acquisition of the two lines resulted in an additional 50 km of excellent subsurface data. With the assistance of the Seismic Contractor BGP, a 3D cross-spread (2.5D survey) was acquired to evaluate potential offset drilling from a recent well (Figure

14a). The cross-spread data were then processed into a small 3D. In order to increase the fold of the 3D, the data was binned at twice the 2D bin size (25 m x 25 m vs. 12.5 m). This resulted in a nominal fold of 48; however, the interpretable area was constrained by the mute offsets and to where the fold contribution was from more than one cross-spread. Despite these limitations, the survey provides great benefit by imaging closures and providing detailed information for possible future offset well drilling (Figure 14b). Given the improvement in imaging and the minimal cost, it is recommended that in cases where a 3D survey is not possible due to time constraints, 3D cross-spreads be considered for targeted imaging.

Figure 14: (a) Cross-spread geometry with red lines representing active 2D lines (sources & receivers) during recording, yellow area representing area of recorded cross-spread data; (b) cross-spread time slice showing good data quality.

Conclusions Poor data areas in Southern Chad were observed to be associated with laterite occurrence, as mapped on vintage soil maps. Where possible, new seismic lines were placed to avoid surface occurrences of crystalline laterite. Based on the test results acquired during the 2013-2014 seismic acquisition program, two different sets of parameters were recommended for acquiring 2D data during 2014-2015 seismic acquisition. In areas with difficult near surface conditions (surface laterites, near surface scatterers), high effort acquisition with aerial receiver arrays, smaller spatial sampling interval, and multiple sweeps were recommended. In areas with good data quality, efficient linear receiver arrays, a larger spatial sampling interval, and single sweep were used. The excellent data quality obtained from the 2014-2015 seismic acquisition program illustrates the benefits of incorporating seismic acquisition tests into regular seismic acquisition to improve quality and reduce costs. It is expected that results from the testing conducted during the 2014-2015 seismic acquisition program will provide further benefits for future 3D programs planned in the area. It has been shown that with ongoing seismic operations, seismic acquisition tests can be incorporated into every

regular seismic acquisition program with very little additional cost into order to continually improve data resolution. Acknowledgements The authors thank Glencore UK Ltd. for providing the data and interpretation, OptiSeis Solutions Ltd. for providing the analysis and seismic test designs, Thrust Belt Imaging for the processing, Larry Cameron, Geoffroy de Vernou and Don Crowell for their assistance in field acquisition and the Ministère de l'Energie et du Pétrole (Chad), and the Société des Hydrocarbures du Tchad. The authors would also like to acknowledge the cooperation of the seismic contractor BGP International (Chad) and the strong performance of BGP management Mr. Ling and Mr. Liang in advancing the application of seismic in Southern Chad. References Audry and Poisot, 1966. Carte Pédologique de Reconnaissance de la République du Tchad: Niellim. ORSTOM. Barbery and Bouteyre, 1967. Carte Pédologique: Moundou (Mission 1959). ORSTOM. Boucquier and Barbery, 1966. Carte Pédologique de Reconnaissance de la République du Tchad: Singako. ORSTOM. Bouteyre, G. 1967. Carte Pédologique: Koumra (Mission 1960). ORSTOM. Marius and Barbery, 1964. Carte Pédologique: (a) Fort Archambault (Mission 1962). ORSTOM, (b) Moussafoyo (Mission 1961). ORSTOM. Meunier, J. 2011. Seismic Acquisition from Yesterday to Tomorrow, 2011 Distinguished Instructor Short Course, Distinguished Instructor Series, No. 14. Society of Exploration Geophysicists. Vermeer, G. 2012. 3D Seismic Survey Design Second Edition, Vol 12 Geophysical Reference Series, Society of Exploration Geophysicists. Vizier and Fromaget, 1969. Carte Pédologique de Reconnaissance de la République du Tchad: (a) Lai. ORSTOM, (b) Fianga. ORSTOM.