Article Type: Research Article Article Citation: T. Kishan Rao, Dr M Shankar Lingam, Manish Prateek, and E. G. Rajan. (2020). SUBSURFACE IMAGING USING GROUND PENETRATING
RADAR FOR UNDERGROUND OIL AND MINERALS EXPLORATION. International Journal of
Engineering Science Technologies, 4(6), 71-79. https://doi.org/10.29121/IJOEST.v4.i6.2020.139 Received Date: 20 November 2020
Accepted Date: 31 November 2020
Keywords: Oil Exploration Ground
Penetrating Radar Vertical Seismic
Profiling An oil well drilling company contractor drills an oil well will as per specific requirement like the depth. It has been estimated that cost to drill an oil well is about $17.01/ Barrels-of-Oil-Equivalent (BOE). More specifically, cost for drilling an onshore well is $13.38/BOE and for drilling an offshore well is $49.54/BOE. Based on some statistics one out of five rigs drilled in an area yields oil recovery. This means $20 million has to be spent for prospecting and locating one oil well. The question that arises now is whether it is worth trying all possibilities of reducing the cost of locating an oil well to $4 million and save $16 million. Well, the research presented in this paper is aimed at showing such a possibility.
1. INTRODUCTIONThree-dimensional digital image processing has numerous applications in the fields of medical imaging, seismology, geomorphology, remote sensing and material sciences. Based on the requirement to solve a problem of interest, one would classify features of three-dimensional images and apply relevant algorithms. For example, the quantitative analysis of geomorphological structures in GPR radargrams and other similar images requires extraction of volumes and shapes of geological constructs like oil reserves, ores and minerals, and encapsulated gems, gold and diamonds present below the surface of earth. In order to do this, there is a need for an aerial survey of the area of interest for locating the land for scanning using ground penetrating radar. For example, drilling an oil well is done by a contractor. The contractor would operate as per the contract terms, like depth of the well and width of the bore. The contractor charges the beneficiary on per day or else per foot rate of drilling. In 2007 it was costing $4 million to drill a well and $3.9 million to drill a gas well. The average cost to drill a well has been $574/foot of depth. The average cost to drill oil or gas well property in USA was $17.01/ The onshore development cost was $13.38/BOE and offshore development cost was $49.54/BOE. Based on some statistics one out of five rigs drilled in an area yields oil recovery. This means $20 million has to be spent for prospecting and locating one oil well. The question that arises now is whether it is worth trying all possibilities of reducing the cost of locating an oil well to $4 million and save $16 million. The research presented in this paper is aimed at showing such a possibility. 1.1. OBJECTIVE OF THE RESEARCHMajor objectives of this intended research are: 1) To develop new methodologies for data acquisition, processing, interpretation 2) To integrate multiple technologies focusing VSP (Vertical Seismic Profiling) methods and appropriate interpretation in order to locate new oil reserves. 3) To make use of VSP (Vertical Seismic Profiling) methodology for prospecting minefield development and management for maximum economic recovery of various minerals and gems. 1.2. SUB-OBJECTIVES OF THIS RESEARCH1) To device methodologies for collecting data from underground of the earth crust. Data acquisition sensors like Synthetic Aperture Radar (SAR) or Ground Penetrating Radar (GPR) and Microwave pulse transmitters could be used to achieve this. Other seismographic techniques could also be used alternatively or in association with SAR/ GPR. 2) Multiple Sensor Fusion is the next part. Data collected from various sensors could be integrated as a data set from which quite a lot of information about the subsurface elements could be interpreted. 3) Fast and reliable algorithms for extracting information from the data set is the major R&D component involved in this research work. 4) To develop a state-of-the-art technology to use the developed tools for detecting oil, minerals and gems reserve in a particular area from which genuine and reliable data set has been collected. 5) However, the material presented in this paper is restricted to the fundamental research carried out as to how to prospect a field of interest and how to obtain a 3D radargram for the purpose of data analysis. 2. METHODOLOGY TO BE USED IN PROSPECTING A FIELD OF INTERESTWhile radar systems work in a free-space environment, GPRs work in media that exhibit electromagnetic properties, which are frequency-dependent. So, the reflected electromagnetic energy shows dispersive effects. Radar systems are used to detect targets at ranges of hundreds of kilometers, but in the case of GPR, targets at ranges of some meters only are detected because of limitation in the transmitted power and attenuation of the reflected signal. Moreover, resolution of GPR is very much limited to centimetres or metres (based on the active frequencies, which are smaller when compared to those used in tactical radar systems). In any case, one has to employ frequencies ranging from tens of MHz to some GHz in order to achieve good spatial resolution of the reflected data. Different GPR configurations are used in land surveys and their choice is made based on different applications like (i) the type of exploration to be made, (ii) the type of targets to be detected, and (iii) the extent of region to be explored. For example, the region to be explored is studied initially by an aerial survey. Fig. 1 shows the image of an area obtained using a drone camera. After visualizing the real field image obtained using a drone camera, a geomorphologist would be able to decide on the fact whether the concerned area is worth to be prospected.
Figure 1: Aerial visualization of an area to be prospected using a drone camera Let us consider an oil field model shown in Fig. 2 as a rectangular / hexagonal pattern.
Figure 2: An oil field hypothetical model Rectangular model is generally preferred to avoid operational difficulties. The rectangular field model is divided into a matrix of 11X9 sub regions. Scanning of a minefield is carried out using a GPS controlled autonomous mobile robot (AMR) system consisting of a ground penetrating radar and image acquisition system. Serpentine scanning is used for prospecting mine fields. Fig. 3 shows the scanning types generally used by an AMR.
Figure 3: Scanning types used by an autonomous mobile robot system Ground Penetration Radar (GPR) is used for excitation. Excitation is applied at appropriate locations, preferably at the centers of the sub divided regions. Fig. 4 shows excitation process as a serpentine scanning of a rectangular subsampled field model.
Figure 4: Excitation process as a serpentine scanning of rectangular field model Area of a sub divided region determines the resolution. The intensity of excitation depends on many factors but mainly on the required depth of penetration. A short duration pulse is applied as excitation and the echo received and registered for further analysis. With reference to the mine field model shown, one would obtain 99 echoes in the form of one-dimensional characteristic signals. The hypothetical excitation points are shown in Fig. 5.
Figure 5: Hypothetical excitation points in the model The received signal from an excitation point both in the analog and discrete form is shown in Fig. 6.
Figure6: Acquired radar echo from under the earth surface in analog and discrete form The sampling of the analog signal is carried out with the Nyquist frequency without causing any loss in the vital information related to characteristic echo due to various sub surface elements. Fig. 7 shows ensemble of 1-D signals (echoes spatially distributed) assumed for the hypothetical field model, its solid model and sectional views.
Figure 7: Ensemble of discrete signals, solid model and its sectional views The ensemble of discrete signals is viewed as a sparsed 3D data. One can dilate this sparsed data in order to get a solid model. Fig. 7 shows the solid model of the sparsed data, elevation and right side full section views of the VSP Stack. VSP refers to Vertical Seismic Profile. VSP is also known as sectional profile. HSP refers to Horizontal Seismic Profile. HSP is also known as depth slice or time slice. VSP stack is generally used to predict oil and other mineral reserves, whereas HSP is used for archaeological survey. Fig. 8 shows a hypothetical VSP and HSP. Figure 8: A sample VSP and HSP Once VSP or HSP is obtained, one can use 3D
image processing procedures to obtain hidden evidence for locating subsurface
components of interest. Discrete lines
of GPR data denote a sectional subsurface view. Multiple lines of GPR data
collected over a specific area are used to obtain relevant three-dimensional
images. Data could be viewed either as three-dimensional blocks, or as
horizontal/vertical slices. Horizontal slices (known as "depth
slices" or "time slices") denote plan view plots separating
specific depths. Time-slicing is a standard practice adopted by archaeologists,
because horizontal patterning mostly indicates many evidences
related to cultural activities. A survey was
made in November 1993 using analog GPR equipment on Nutubaru
Kofun burial site in Shintomi Machi, Ishikawa, Japan
and 3D depth slices obtained. Fig. 9 shows a large round anomaly found at a depth
as reflections from the moat of a destroyed burial mound dating from the 6-7th
century AD. The full description of the history of this mound could be seen in
various literature. The authors acknowledge Geophysical Archaeometry Laboratory
Inc., 20014 Gypsy Ln Woodland Hills, California 91364 for providing details
about the visual material related to GPR applications presented in this paper. For
details visit their web site: http://www.gpr-survey.com/.
Figure 9: HSP slices obtained using a GPR at a burial site in Japan Figure 10
shows actual HSP slices and the corresponding image of the excavated burial
site in Japan.
Figure 10: Fourteen depth slices of the HSP obtained
using a GPR at a burial site in Japan 3. GROUND PENETRATING RADAR AND RADARGRAM ACQUISITIONAn archaeological site is surveyed by a mobile GPR at a finite number of locations at the site denoted by x and y, and radar excitation pulses fired through the ground surface. The reflected echoes due to individual pulses representing the depth of 3rd dimension, say z are collected and recorded from which the 3D radargram image is developed. In order to make such 3D images, various signal processing techniques are used. Signal processing operations are certain mathematical models and they carry out noise filtering and signal integration on raw radar pulse data. Some of the basic GPR data processing operations are briefly explained below. Post Processing Gain Improvement Usually, GPR data are 16-bit discrete sequences. The discrete data samples obtained from GPR echoes vary from -32768 to +32767. Amplitudes of the reflected echoes from just below ground surface to an intended depth would vary from a maximum to a minimum. The resistance offered to echoes that travel from a depth to the surface would be maximum so that the echoes from various layers of earth would be weaker and weaker depending on the depth. This calls for gain improvement, possibly an automatic gain control has to be used in order to study the characteristic reflections from various scatterers from under the ground. DC Drift Removal The gain corrected raw radargrams would often be corrupted by different kinds of noise and this would shift the DC base line above ‘0’. In such a case, one would use appropriate filters, like wiener filter, anisotropic diffusion filter, log-normal filter, depending on the type of noise like additive or multiplicative noise or speckle noise, present in the radargram. Since the base line shifted radargram will not appear as true visual model of the actual subsurface components, DC restoration is done on the radargram with the help of denoising filters. Background Removal The echoes obtained from various depths not only due to scatters of specific interest but also due to soil, rock and other scatterers of no interest. Such echoes are called background noise. The echoes from scatterers are interest are called signatures. Background removed signatures would yield a meaningful visual presentation due to scatterers of interest. Migration Moving a GPR along a profile line would cause the profile of echoes from a particular scatterer to look like an inverted hyperbola. This is due to the fact that the GPR antenna would not only receive echoes from a scatterer when the antenna is just above it but also receive echoes from the same scatterer even after the GPR moving away from it. In addition, noise would also corrupt the presentation of the inverted hyperbola. In such a case, an operation called migration is carried out as a signal conditioning measure. The process of migration is caried out by fitting a hyperbola corresponding to a scatterer by adding up all energy values along the hyperbola and keeping this addition right at the top of the inverted hyperbola. This is done for every point in the radargram. The net result is to break the hyperbolas to smaller point scatterers. The next step after migration, the radargram should be processed to obtain a meaningful image from this data. Time slice presentation is the most useful image for archaeological applications. Time slices or HSPs represent maps of amplitudes of recorded reflections across a site at a specified time or depth. One could follow the following technique to create the most meaningful images for archaeological applications.
Time Slice Processing
of 3-D images using cellular logic methods There is another novel paradigm called ‘Cellular Logic Array Processing (CLAP)’ introduced by E. G. Rajan, in which one can develop fast algorithms for processing digital images. Cellular Logic Array Processing is a logico-mathematical framework developed using the fundamental notions of Markov’s ‘Normal Algorithms’ and von Neumann’s ‘Cellular Automata’. Normal algorithm is an idealized serial processor like a Turing machine and a cellular automaton is an idealized parallel processor. CLAP is a logical technique of realizing cellular automata using normal algorithms. Expected
Research Impact The research output would open up new avenues for developing a prototype product and whose future versions would be used in various 3-D image processing applications like ‘Medical Image Processing’, ‘Mineral Exploration’, ‘Archeological Excavations and ‘Defense Applications’. Some of the imminent benefits of this research are envisaged as outlined below. Societal
benefits ·
Radiologists,
medical professionals can identify various features from 3-D images for
diagnostic purposes. ·
Prediction
of damaged tissues/organs in less time without going for surgery or biochemical
test. ·
Useful
for researchers in analyzing the 3-D structures in various fields like
oncology, geology, zoology, aviation, geophysics, and protein modeling etc. ·
It
is useful for detecting un-exploded landmines, where layer wise 2-D images of particular land are obtained by Ground Penetrating Radar
devices. ·
It
is useful for knowing the position of archeological objects under the ground so
that objects are not damaged while excavating from the ground. Economic
benefits ·
Oil
companies spend lots of money in digging wells to explore the presence of oil.
The 2-D image of the layers of the earth can be obtained by using ground
penetrating radar (GPR) device. The sequence of these 2-D images can be
reconstructed to form a 3-D image called Vertical Seismic Profile (VSP) and
such 3-D images can be processed to obtain information about the presence of
oil or other elements of interest. By using this technique oil companies can
have a priori knowledge about a site and then go for a rig which means saving
of money. · Similarly, medical professionals and other people like 3-D film producers who make use of 3-D image processing and analysis can get their jobs done effectively, reliably and with reduced cost. SOURCES OF FUNDINGThis research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. CONFLICT OF INTERESTThe author have declared that no competing interests exist. ACKNOWLEDGMENTNone. REFERENCES[1] Jean Serra; Cube, cube-octahedron or rhomb-dodecahedron as bases for 3-D shape descriptions, Advances in visual form analysis, World Scientific 1997, 502-519. [2] Wuthrich, C.A. and Stucki, P.; An algorithm comparison between square- and hexagonal based grids; Graphical Models and Image Processing, 53(4), 324–339,1991. [3] Reinhard Klette and Azriel Rosenfeld; Digital Geometry: Geometric methods for picture analysis; Elsevier, 2004. [4] B. Nagy; Geometry of Neighborhood sequences in hexagonal grid; Discrete Geometry of computer imagery; LNCS-4245, Springer. [5] Tristan Roussillon, Laure Tougne, and Isabelle Sivignon; What Does Digital Straightness talk about Digital Convexity? [6] Edward Angel, “Interactive Computer Graphics- a top-down approach with OpenGL”, Second Edition, Addison Wesley, 2000. [7] S.W. Zucker, R.A. Hummel, “A Three-Dimensional Edge Operator”, IEEE Trans. On PAMI, Vol. 3, May 1981. [8] Jürgen, H., Manner, R., Knittel, G., and Strasser, W. (1995). Three Architectures for Volume Rendering. International Journal of the Eurographics Association, 14, 111-122 [9] Dachille, F. (1997). Volume Visualization Algorithms and Architectures. Research Proficiency Examination, SUNY at Stony Brook [10] Günther, T., Poliwoda, C., Reinhart, C., Hesser, J., and Manner, J. (1994). VIRIM: A Massively Parallel Processor for Real-Time Volume Visualization in Medicine. Proceedings of the 9th Eurographics Hardware Workshop, 103-108 [11] Boer, M.De, Gröpl, A., Hesser, J., and Manner, R. (1996). Latency-and Hazard-free Volume Memory Architecture for Direct Volume Rendering. Eurographics Workshop on Graphics Hardware, 109-119 [12] Swan, J.E. (1997). Object Order Rendering of Discrete Objects. PhD. Thesis, Department of Computer and Information Science, The Ohio State University [13] Yagel, R. (1996). Classification and Survey of Algorithms for Volume Viewing. SIGGRAPH tutorial notes (course 34) [14] Law, A. (1996). Exploiting Coherency in Parallel Algorithms for Volume Rendering. PhD. Thesis, Department of Computer and Information Science, The Ohio State University [15] Ray, H., Pfister, H., Silver, D., and Cook, T.A. (1999). Ray Casting Architectures for Volume Visualization. IEEE Transactions on Visualization and Computer Graphics, 5(3), 210-233 [16] Yagel, R. (1996). Towards Real Time Volume Rendering. Proceedings of GRAPHICON' 96, 230-241 [17] Kaufman, A.E. (1994). Voxels as a Computational Representation of Geometry. in The Computational Representation of Geometry SIGGRAPH '94 Course Notes [18] Lacroute, P., and Levoy, M. (1994). Fast Volume Rendering Using a Shear-Warp Factorization of the Viewing Transform. Computer Graphics Proceedings Annual Conference Series ACM SIGGRAPH, 451-458. [19] Sutherland, I.E., Sproull, R.F., and Schumaker, R.A. (1974) A Characterization of Ten Hidden Surface Algorithms. ACM Computing Surveys, 6(1), 1-55 [20] Roberts, J.C. (1993). An Overview of Rendering from Volume Data including Surface and Volume Rendering. Technical Report 13-93*, University of Kent, Computing Laboratory, Canterbury, UK. [21] Frieder, G., Gordon, D., and Reynolds, R.A. (1985). Back-to-Front Display of Voxel-Based Objects. IEEE Computer Graphics and Applications, 5(1), 52-60 [22] Westover, A.L. (1991). Splatting: A Parallel Feed-Forward Volume Rendering Algorithm. Ph.D. Dissertation, Department of Computer Science, The University of North Carolina at Chapel Hill [23] Zwicker, M., Pfister., H., Baar, J.B. and Gross M. (2001). Surface Splatting. In Computer Graphics SIGGRAPH 2001 Proceedings, 371-378 [24] Nulkar, M., and Mueller, K. (2001). Splatting with Shadows. International Workshop on Volume Graphics 2001,35-50 [25] J. Krüger and R. Westermann, Acceleration Techniques for GPU-based Volume Rendering, Proceedings of the 14th IEEE Visualization 2003 (VIS'03), 38-43. [26] Markus Hadwiger, Joe M. Kniss, Christ of Rezk-salama, Daniel Weiskopf, and Klaus Engel. Real time Volume Graphics. A. K. Peters, Ltd., USA, 2006. [27] Goodman, D., Nishimura, Y., Hongo, H and Noriaki N., 2006. Correcting for topography and the tilt of the GPR antenna, Archaeological Prospection, 13: 157-161. [28] Goodman, D., Y. Nishimura, and J. D. Rogers, 1995. GPR time slices in archaeological prospection: Archaeological Prospection, 2:85-89. [29] Goodman, D., 1994. Ground-penetrating radar simulation in engineering and archaeology: GEOPHYSICS, 59:224-232. Advisory Committee Members and Co-authors [1] Dr. Amit Agarwal, Director, APJ Abdul Kalam, Institute of Technology, Dehradun, India [2] Dr. Michael Patrick Coyle, Chief Executive Officer, Avatar MedVision US LLC, NC, USA [3] Sathya Govindarajan, Director, Pentagram Research Centre Private Limited, Hyderabad, India [4] Prashanthi Govindarajan, Director, Pentagram Research Centre Private Limited, Hyderabad, India [5] Yashaswi Vemuganti, Consultant, Avatar MedVision US LLC, NC, USA [6] Dr. Jean Claude Perez, IBM European Research Center on Artificial Intelligence, France [7] Dr. Hindupur Rajasimha, Chief Engineer, Indian Space Research Organization and IDBI (Retd.)
This work is licensed under a: Creative Commons Attribution 4.0 International License © IJOEST 2016-2020. All Rights Reserved. |