Fast stochastic simulation of patterns using a multi-scale search

Size: px
Start display at page:

Download "Fast stochastic simulation of patterns using a multi-scale search"

Transcription

1 1 Fast stochastic simulation of patterns using a multi-scale search Pejman Tahmasebi and Jef Caers Department of Energy Resources Engineering Stanford University Abstract Pattern-based spatial modeling relies on the pattern as the basic modeling component for generating geostatistical realizations. These methodologies recognize that working with the unit of a pattern aids both data conditioning (an advantage shared with pixel-based methods) and better pattern reproduction (an advantage shared with Boolean models). Since the first paper on conditional simulation with patterns by Arpat and Caers (simpat, 2005, 2007), a series of methodologies for simulation have been developed attempting to overcome the main limitation of simpat: CPU time and variability of the generated realizations. In this paper we extend on the ideas of ccsim, a method for pattern simulation relying on the cross-correlation as measure for pattern similarity. The extension lies on the use of a multi-scale representation of the training image along a pattern projection strategy that is markedly different from the traditional multi-grid employed in current methodologies. In this multi-scale representation, we transform the high-resolution training image into a pyramid of consecutively up-gridded views of the same training image. This pyramid allows for rapid search of patterns to interlock with the previously simulated patterns. A second advantage of this multi-scale view lies in data conditioning. Using multi-million cell examples with sparse as well as dense datasets, we illustrate how this method can simulate such large grids in a matter of seconds. Introduction Since more than two decades from its inception, research in multiple-point geostatistics has generated a prolific development in new algorithms whose aim is to capture and reproduce patterns from training images, anchored to local hard and soft data. Regardless of these developments, the wide-spread application in practice is still lacking, even in reservoir modeling, where most of these techniques have received considerable attention. We conjecture that, next to the availability of training images (see discussion in Honarkhah and

2 2 Caers, 2012), two main reasons lie at the origin of this: quality of pattern reproduction and CPU time. In this paper, we present a method that can simulate multi-million cell grids in a matter of second, and, provide a quantum improvement in terms of pattern reproduction over existing algorithms, while maintaining accuracy in hard data conditioning. In his original and seminal algorithm, Strebelle (2002) used a search tree in a single normal equation simulation (snesim) to accelerate the calculation of local conditional probabilities rendering the application of multiple-point geostatistics practically possible. snesim has considerable memory requirements for simulating multiple-facies, although this issue has been addressed in for example Straubhaar et al. (2011). However, these solutions do not address the difficulty in reproducing thin features or very low proportion patterns (curvi-linear shale barriers, fractures etc). Pattern-based methods, proposed originally by Arpat and Caers (2005, 2007), aim to overcome these RAM limitations, as well as improve pattern reproduction quality. These methods basically consider stochastic simulation as a randomized puzzle where patterns are puzzle pieces that need to interlock (and be constrained to data) to generate a realization. In the same context, Zhang et al. (2006, filtersim) provided a filter-based selection of patterns to speed-up the time-consuming search of simpat, however at the cost of lesser pattern reproduction. Honarkhah and Caers (2010, dispat) relied on a distance-based pattern clustering to rapidly search for patterns without the approximations made in filtersim. Tahmasebi et al. (2012) recognized that using a cross-correlation as a distance as well as simulating along a raster path (instead of random path) provided great improvements over filtersim. In this special issue, this ccsim (cross-correlation simulation) is indeed quantitatively proven to perform better than dispat (see Tan et al, this issue). As an extension, Tahmasebi and Sahimi (2012) used a frequency domain to provide more acceleration on large grids. In this paper, we present a new pattern-based method that achieves two goals 1) extremely fast CPU performance and 2) improved pattern reproduction (an accompanying paper by Tan et al, this issue, goes in great detail to provide a quantitative assessment of this statement). Our proposed method is cast on the ccsim framework, an algorithm that will be reviewed first. To accelerate ccsim, we propose a multi-scale method for searching and pasting patterns in the training image. Based again on a multi-scale representation of the training image and simulation grid, a new method for hard data conditioning is established. A comparison (on CPU and pattern reproduction) by means of extensive, categorical, continuous, conditional and unconditional simulation on multi-million cell grid is performed between the existing ccsim and the new ms-ccsim (ms=multi-scale).

3 3 A review of cross-correlation-based simulation (ccsim) Unconditional ccsim ccsim extends on the idea of simpat in two ways: 1) the similarity between any pattern and the existing data-event is based on a cross-correlation over a smaller overlap region and 2) a raster path (similar to Daly, 2005; Stien & Kolbjornsen, 2011) is used instead of a random path. Figure 1 and 2 explain the basic concepts of the algorithm. Following a raster path starting at the origin, Figure 1A, a random pattern is selected from the training image, see Figure 1E. Next, the template (which is always square) is shifted to the right, Figure 1B, leaving a small area of overlap (size h, see also Figure 2). It is the overlapping area that, just as in a real puzzle, allows deciding on what the best interlocking pattern is. The criterion for distance used is the crosscorrelation between the first pattern pat 1 and any other pattern pat k in the training image summed over the overlap areas = d( pat, pat ) pat pat (1) 1 k 1, i k, j i A j A A pattern is a vector of grid cell values pat = { pat, pat, K, pat } n T = number of pixels in the template T (2) k k,1 k,2 k, n T A threshold t allows selecting any pattern in the training image that has a distance less than t in Eq. (1) (otherwise one would select the pattern in the training image right next to pat 1 ). Notice that the calculation in Eq (1) is over the overlap area A only, not over the entire pattern (as in dispat), which provides a considerable speed-up. After the first line is completed, most of the pattern cross-correlation calculations involve the geometry shown in Figure 2A, where the shaded area is used to find interlocking patterns (tantamount to finding puzzle pieces with only a bottom and left interlocking component). The above algorithm constitutes the basic idea for unconditional simulation. The randomization is mostly due to the selection of the first pattern and the threshold t. The algorithm requires specification of three parameters: the size of the template (n T ), the size of the overlap area A (h) and the threshold (t). The size of the template is determined in the same way as in dispat, namely using the so-called elbow plot (pattern entropy versus template size). Tahmasebi and Sahimi (2012) found through sensitivity analysis that a reasonable choice is to take h between 1/5th and 1/6th of the template size T. In similar vein, a reasonable choice is for t to be 20% larger than the minimum distance in Eq (1) or

4 4 t = 1.2 min patm, ipatk, j k, m i A j A (3) Figure 1: concept of raster path and overlap region. Figure 2: (A) configuration of template, overlap area A, (B) when hard data are present, (C) splitting of the template.

5 5 Conditional ccsim For conditional simulation, a two-step search similar to simpat and dispat is used. Consider the situation in Figure 2B. First, one collects all the patterns k into a set of patterns: set = { pat pat t k : d( pat previous, pat < k) < t} (4) namely, find the set of patterns that interlocks the previously simulated patterns in Figure 2B. Then, for any given hard data within the template T (see Figure 2B), one selects a pattern from set pat<t those that match the hard data in the data event dev T. It is likely (in particular with dense hard data and inconsistent training image) that no such pattern can be found. In that case, the template T is split into a smaller template, see Figure 2C. The two-step operation is now repeated with this smaller template (similar to dropping far away hard data in snesim). This operation is continued until a matching pattern is found. Note that conditioning is different from Markov mesh models (MMM, see Kjønsberg & Kolbjørnsen, 2008), where an additional kriging is needed to account for any hard data ahead in the raster path. In MMM one simulates pixel values, not patterns. Simulating with patterns allows the algorithm to look for hard data ahead of the raster path. The newly proposed multi-scale approach enhances this looking-ahead for hard data even more, as explained next. Multi-scale modeling Multi-grid vs multi-scale simulation In many MPS algorithms (e.g. snesim, dispat), a multi-grid simulation is used to enhance longrange pattern reproduction. The multi-grid idea starts from a sparse but coarse grid and then successively simulates on progressively finer grids, see Figure 3. It was originally introduced in the context of covariance-based models (Tran, 1994). Later, the same idea was introduced to enhance the reproduction of higher-order statistics borrowed from training images (Caers and Journel, 1998). Simulating on multiple-grids is not without its issues. First, since the first grid to be simulated is coarse and sparse, there is the issue of how to deal with irregularly spaced or clustered data. Some method of data relocation or allocation needs to be introduced and this may result in poor conditioning (Honarkhah, 2011), locally biased pattern reproduction or reduction in uncertainty over multiple realizations (Caers, 2012). The second issue is that of global pattern reproduction. During coarse grid simulation, patterns are constructed from a much sparser template. This sparseness is a source of randomness in the patterns, which may induce

6 6 randomness in the generated realizations at the coarsest grid level that is maintained (by construction) at the finer scale grid simulation, leading to poor pattern reproduction quality. In this paper we propose using an entirely different approach based on resolution theory, common to many methods of mimicking the human visual system (a similar approach was presented in Honarkhah, 2011). Figure 3: simulating on a multi-grid. Starting from a coarse grid and sparse template, the simulation grid is consecutively refined until the final resolution. Multi-Scale Theory Our visual system makes measurements by integrating information around us, either small objects or large scenes. The eye does not limit itself to small or large objects. It has the flexibility to gauge its surroundings without any prior information on what to expect, whether the object is big or small, whether it is square or a line. The visual front-end of the eye has been regarded as a multi-scale geometry engine" (Koenderink, 1984). When an observation is made in space, all information is transmitted to the brain, as not simple one single image, but as a stack at different scales. This is the basic principle in scale-space theory that aims at explaining how the visual front-end performs (Lindeberg, 1994). We propose a multi-scale approach based

7 7 on the psycho-visual model of human optical system. However, pattern-based simulation algorithms are not able to handle all resolutions simultaneously due to the presence of local hard data. Instead, we propose to maintain separate pattern resolutions in the training image. In the same way as humans analyze a structure by first blurring the features to get a big picture representation of the model, and then focusing on details and fine-scale patterns, we search the training image starting from the coarsest resolution and follow a top-down approach. In the next section these principles are implements in the ccsim algorithm. Unconditional multi-scale cross-correlation-based simulation (ms-ccsim) In a multi-scale approach, we first up-grid the training image into a pyramid of multi-resolution views of the same training, see Figure 4. The finest resolution training image is the original training image and termed ti 0. We use bi-cubic interpolation (Lekien & Marsden, 2005) to obtain values for the lower-resolution training images ti 1,..ti G-1, where G is the total number of resolutions considered. The template size for up-gridding remains the same for all resolutions (in 2D 4x4, in 3D 4x4x1 to maintain vertical variability). In case of binary variables, Otsu s thresholding method (Otsu, 1979) is used to turn the continuous valued bi-cubic interpolations back to binary variables (categorical values are basically a vector of binary variables), see Figure 5. Figure 4: procedure for up-gridding by tri-cubic interpolation.

8 8 Figure 5: example application of up-gridding a binary training image using tri-cubic interpolation and Otsu s thresholding. The unconditional ms-ccsim proceeds similarly to the unconditional ccsim, except that the training image as well as overlap area A is now upgridded. Based on the multi-scale representation of the training image, we propose a multi-scale patterns search as depicted in Figure 6. One starts from the lowest resolution training image tig-1 and finds a pattern in tig-1 that interlocks the previous simulated patterns at that resolution. Next, the same location in the next scale g-2 is determined and a search window (red box in Figure 6) is used having the following size Search box size S = 2 ( template size T for grid g-2 Overlap h for grid g-2 ) Note that the template size T as well as overlap size h doubles when going from a resolution g-i to a resolution g-i-1. The idea of using this search box is that there is no longer a need to search for a matching pattern over the entire training image ti g-2, the search is limited to this smaller neighborhood. This is continued until the final resolution ti 0. The pattern found at this resolution is then pasted on the simulation grid.

9 9 Figure 6: principle of the multi-scale search: starting from a coarse resolution version of the training image, a matching pattern is search top-down in the pyramid. Conditional ms-ccsim - Categorical variables In the traditional multi-grid approach, Strebelle (2000) uses a data relocation approach, where the hard data are relocated to their closest grid node in the multi-grid realizations. This ensures that hard data are taken into account during coarse grid simulations, and as a result, the finer grids are informed about the larger scale statistical dependencies of hard data. This works for simple cases, but with locally dense data, conflicts arise, and as a result, hard data are dropped at coarse grid simulations leading to poor conditioning. In this paper, we present a method for hard data conditioning that is adapted to the multi-scale representation of the training image presented above. First, we would like to stipulate very clearly the nature of a multi-resolution grid g. As in any geostatistical approach we basically work with a corner-point representation for each multi-resolution grid (see Deutsch and Journel, 1992), meaning that at every corner of a grid, a cell of certain volume (often the same as the volume of any hard data, which can be tiny) is located, regardless of the resolution g, see Figure 7. The volume v of this cell is the same for all resolutions g.

10 10 Figure 7: Definition of the geostatistical grid for simulating a property with volume v on the primal mesh. Figure 8: various possible configurations of hard data within the dual mesh. (A) simplest case of not more than one hard data per dual mesh grid cell, (B) case requiring re-assigning hard data based on distance to the nearest location. (C) case where multiple hard data cannot be assigned to the nearest location: in that case we assign that category with the highest frequency. Consider now a multi-resolution grid g with a number of possible configurations of hard data. The simplest situation arises in Figure 8A, where each cell in a dual multi-resolution mesh g contains no more than one hard datum. In that case any hard data (if present) is assigned to the

11 11 geostatistical cell in the center. If more than one hard datum is present in a single cell of the dual mesh, then hard data are assigned based on the distance between the geostatistical cell location and that hard datum. If however, too many hard data are present and not all hard data can be assigned, then a frequency distribution of the remaining hard data is calculated and the geostatistical cell is assigned the category with highest frequency. In fact the latter operation is tantamount to up-gridding the hard data to the multi-resolution grid. - Continuous variables The pattern conditioning in the case of continuous-valued training image is considerably more challenging than in the binary case, simply because one rarely finds an exact match between a training image pattern and a hard data event. In addition, when applying the multi-scale approach, the original training image is averaged into progressively smoother version of that training image. Hence, it becomes even less likely, at the coarse resolutions, to find a pattern that matches the hard data-event. To alleviate this problem, we apply a histogram transform, on each multi-resolution grid of the original training image, the target histogram being the original training image histogram. The continuous data are then categorized into bins and the data conditioning follows the same methodology as for the categorical case. Examples Unconditional simulation We present various examples of training images with each three unconditional realizations generated from them. Some of the parameters used are summarized in Table 1. Table 1: Summary of template size, overlap size, number of multi-scale levels and CPU time for different examples in this paper. Example Template Number of CPU time (s) h for g=2 size for g=2 resolutions g (Matlab) 2D binary channel D cobbles D channel with crevasse D continuous D binary fracture network D tank data

12 12 Figure 9 shows a 1000x1000 binary channel case: the difficulty lies in the reproduction of long connected sinuous channels. Figure 10 shows a gray-level training image of cobbles: the difficulty lies in reproducing closed curved features. Figure 11 shows a 3D channel training image with crevasse splays: the difficulty lies in creating connected channels in 3D. Figure 12 shows a 3D training image generated with sequential Gaussian simulation (using a Gaussian variogram): the difficulty lies in the smooth spatial continuity that is difficult to reproduce with pattern-based method (filtersim and dispat realization often display an artificial patchiness related to the template size). Figure 13 shows a training image extracted from a tank experiments (Paola et al. 2001; Connell et al., 2012). Tank experiments are laboratory scale experiment simulating sediment deposition in realistic circumstances. After the experiment is completed, the box of sediments is cut into consecutive slices of which photographs are taken. One such photograph is show in Figure 13 (top). From the series of consecutive slices we then constitute a single 3D training image for testing geostatistical algorithms. The patterns seen in these experiments are analogs of real depositional features (notice the scour fill in Figure 13), hence can provide excellent test cases for geostatistical algorithms: the difficulty here lies in reproducing complex sedimentary features represented by a continuous variable. In this particular case we see an excellent reproduction of scour features in Figure 14. Figure 9: 2D binary channel training image on a grid with three unconditional realizations.

13 13 Figure 10: cobble stone training image on a grid with three unconditional realizations. Figure 11: 3D unconditional simulation on a complex 3-facies system.

14 14 Figure 12: Three unconditional realizations for the shown continuous 3D training image. Figure 13: Extracted training image from tank data.

15 15 Figure 14: Three realization by using tank experimental data to construct a training image on a grid. Conditional simulation Figure 15 describes a case for testing the conditioning accuracy of the ms-ccsim. A simple case with two hard data (one channel datum and one background datum) is shown. In this case we can perform rejection sampling, meaning: we perform an unconditional ms-ccsim realization and accept it as it matches the two data. We generate 100 realizations using this rejection sampler. We also generate 100 realizations using the conditional ms-ccsim, of which three are shown in Figure 15. We calculate the ensemble average of both sets of realizations. The ensemble average of the rejection sampler compares well with the conditional method.

16 16 Figure 15: Conditional simulation with two closes hard data. The result of rejection sampling and ensemble average of 100 realizations in a zoom-in view are shown. Using the same training image we set up a case with dense conditioning data: first we generate an unconditional realization and extract a dense data set. Then we generate 100 conditional realizations as shown in Figure 16. The ensemble average shows that conditioning works without creating discontinuities near the data. Figure 17 shows a case with dense continuous data (similar to the cases used in Dimitrakopoulos et al, 2010). The ensemble average shows reasonable conditioning but with some discontinuity in the channel geometry (see enclosed circle). Conditional continuous simulation remains a problem for pattern-based methods as it is difficult to interlock continuous puzzle pieces while at the same time constraining to data.

17 17 Figure 16: Conditional simulation with dense hard data. Reference image which is used for hard data picking, three different realizations, and ensemble average of 100 realizations are shown. Figure 17: Conditional simulation with continuous training image. The hard data is selected in a regular grid pattern from a reference image. One realization and ensemble average are shown.

18 18 Figure 18 and 19 show two challenging cases of conditioning to thin features. Given the complexity of the patterns, the conditioning data as well as the size of the models, standard MPS algorithms fail to produce meaningful results. In Figure 18, we have a training image generated using an ant tracking algorithm on seismic data (Pederson et al. 2002). The result is a delineation of lineaments (faults/fractured zones) that are deemed relevant for an area where no such seismic data is available. In this area however, wells have been drilled and fractured zones interpreted from well-bore an imaging data, as shown in Figure 18 (red color indicated presence of fractures, blue indicated absence). We generate 50 conditional realization of which one is shown in Figure 18. Figure 18: Fracture training image generated using an ant tracking algorithm on seismic data (upper left), hard data (upper right) and one realization (down). To assess the conditioning accuracy, we calculate the ensemble average, shown in Figure 19. A high probability indicated presence of fractures. The zoom shows clearly that the conditional simulation extrapolate from the interpreted fracture zones using the pattern of the training image in Figure 18. However, we can still observe some patchiness in the ensemble average related to the template size; hence some room for improvement is still available. Figure 20

19 19 shows a process-based model of a sand/shale sequence consisting of shale drapes. In reservoirs such shale drapes may compartmentalize the system and affect performance considerably. Figure 20 shows that these shale drapes have a complex 3D geometry that is impossible to reproduce with existing MPS techniques. A single conditional model, conditioned to the 8 wells is shown in Figure 20. The shale drapes intersect at the location where shale is presence in the well, with reasonable, but not perfect, reproduction of the shale drape continuity. Figure 19: Ensemble average of 50 realizations on a grid. Algorithm performance In this section the overall CPU performance of ms-ccsim and ccsim is compared. First, Table 1 shows absolute CPU times for the ms-ccsim algorithm for the cases shown above. All models are generated with less than 1 minute CPU, some only a few seconds, even for multi-million cell models. This performance is several orders of magnitude faster than any existing MPS code. Next, the impact of the multi-scale approach is illustrated in Figure 21. According to this analysis, it is clear that by increasing the size of training image and simulation grid, the superior performance of ms-ccsim through multi-scale search becomes evident.

20 20 Figure 20: A process-based model of a sand/shale sequence consisting of shale drapes (top), hard data (left down) and one conditional realization (right down). Conclusions There is natural logic to use pattern-based approaches for simulating large grids. In multimillion cell models, it is simply inefficient to simulate every single grid cell one at a time whether using a random or raster path. The result is that pixel-based models are too CPU demanding for such cases and secondly that the pattern-reproduction is unacceptably poor. In this paper we build further on the idea of simulating patterns, more specifically on the ccsim algorithm. The main contribution lies in using a multi-scale search for finding a pattern interlocking optimally with previously simulating patterns along a raster path. The multi-scale search provides two main advantages: it speeds up the algorithm such that multi-million models can be simulated in seconds and it improves conditioning to hard data. A large set of complex example of diverse nature testifies of the ability of the new ms-ccsim algorithm. Our further work will focus on improving the pattern reproduction for continuous cases as well as investigate ways of conditioning to secondary data.

21 21 Figure 21: CPU-time comparison for (a) 2D examples and (b) 3D examples. Acknowledgement We would like to thank Dr. Tomomi Yamada of JAPEX for donating the datasets of Figure 18. We would like to thank Dr. Hongmei Li of ExxonMobil for donating the shale drape dataset of Figure 20. We would also like to thank Dr. Chris Paola of the University of Minnesota for the tank experiment data and the many interesting discussions about the use of geostatistics in modeling sedimentary systems. References Arpat, G.B., and Caers, J., 2005, A multiple-scale, pattern-based approach to sequential simulation, Geostatistics Banff 2004, Quantitative Geology and Geostatistics, 14(1): Arpat, G.B., and Caers, J., 2007, Stochastic simulation with patterns, Mathematical Geology, 39(2): Bar-Joseph, Z., El-Yaniv, R., Lischinski, D., and Werman, M., (2001, Texture mixing and texture movie synthesis using statistical learning, IEEE Transaction on Visual and Computer Graphic, 7(2): Caers, J., and Journel, A.G., 1998, Stochastic reservoir simulation using neural networks trained on outcrop data, In: SPE Annual Technical Conference and Exhibition. Society of Petroleum Engineers, New Orleans, LA, pp Connel, S.D., Wonsuck, K., Smith, G. and Paola, C., 2012, Stratigraphic Architecture of An Experimental Basin With

22 22 Interacting Drainages, Journal of Sedimentary Research May 2012 vol. 82 no Daly, C., 2005, Higher order models using entropy, markov random fields and sequential simulation, In: Leuangthong, O., Deutsch, C. V. (Eds.), Geostatistics Banff Vol. 14 of Quantitative Geology and Geostatistics, Springer Netherlands, pp Ebert, D.S., Musgrave, F.K., Peachey, D., Perlin, K., and Worley, S., 2003, Texturing & Modeling: A Procedural Approach, Morgan Kaufmann Publishers. Honarkhah, M., 2011, Stochastic simulation of patterns using distance-based pattern modeling PhD dissertation, Stanford University. Honarkhah, M., and Caers, J., 2010, Stochastic simulation of patterns using distance-based pattern modeling, Mathematical Geosciences, 42: Honarkhah, M., and Caers, J., 2012, Direct pattern-based simulation of non-stationary geostatistical models, Mathematical Geosciences 44(6): Kjønsberg, H., and Kolbjørnsen, O., 2008, Markov mesh simulations with data conditioning through indicator kriging, In: Proceedings of the Eighth International Geostatistics Congress, Santiago, Chile. Koenderink J (1984) The structure of images. Biological Cybernetics 50: Lekien, F., and Marsden, J., 2005, Tricubic tnterpolation in three dimensions, Journal of Numerical Meth and Eng 63: Lindeberg, T., 1994, Scale-space theory in computer vision, Kluwer Academic Publishers, Norwell, MA, USA. Otsu, N., 1979, A threshold selection method from gray level histograms, IEEE Transaction Systems, Man and Cybernetics 9: Paola, C., Mullin, J., Ellis, C., Mohrig, D.C., Swenson, J.B., Parker, G.S., Hickson, T., Heller, P.L., Pratson, L., Syvitski, J. and Sheets, B.N., 2001, Strong Experimental stratigraphy, GSA Today 11(7):4 9. Pedersen, S.I., Randen, T., Sønneland, L., Steen, Ø., 2002, Automatic fault extraction using artificial ants, Expanded Abstract, Int Mtg, Society of Exploration Geophysics, Stien, M., and Kolbjørnsen, O., 2011, Facies modeling using a Markov mesh model specification, Mathematical Geosciences, 43: Straubhaar, J., Renard, P., Mariethoz, G., Froidevaux, R., and Besson, O., 2011, An improved parallel multiple-point algorithm using a list approach, Mathematical Geosciences, 43: Strebelle, S., 2002, Conditional simulation of complex geological structures using multiple-point statistics, Mathematical Geology, 34(1):1-22. Tahmasebi, P., Hezarkhani, A., Sahimi, M., 2012, Multiple-point geostatistical modeling based on the crosscorrelation functions, Computational Geosciences 16(3): Tahmasebi, P., and Sahimi, M., 2012, Fourier acceleration of geostatistical simulations based on cross-correlation analysis, Computers & Geosciences (Submitted) Tran, T.T., 1994, Improving variogram reproduction on dense simulation grids, Computers & Geosciences 20(7-8):

Simulating Geological Structures Based on Training Images and Pattern Classifications

Simulating Geological Structures Based on Training Images and Pattern Classifications Simulating Geological Structures Based on Training Images and Pattern Classifications P. Switzer, T. Zhang, A. Journel Department of Geological and Environmental Sciences Stanford University CA, 9435,

More information

Fast FILTERSIM Simulation with Score-based Distance Function

Fast FILTERSIM Simulation with Score-based Distance Function Fast FILTERSIM Simulation with Score-based Distance Function Jianbing Wu (1), André G. Journel (1) and Tuanfeng Zhang (2) (1) Department of Energy Resources Engineering, Stanford, CA (2) Schlumberger Doll

More information

Multiple Point Statistics with Multiple Training Images

Multiple Point Statistics with Multiple Training Images Multiple Point Statistics with Multiple Training Images Daniel A. Silva and Clayton V. Deutsch Abstract Characterization of complex geological features and patterns has been one of the main tasks of geostatistics.

More information

Short Note: Some Implementation Aspects of Multiple-Point Simulation

Short Note: Some Implementation Aspects of Multiple-Point Simulation Short Note: Some Implementation Aspects of Multiple-Point Simulation Steven Lyster 1, Clayton V. Deutsch 1, and Julián M. Ortiz 2 1 Department of Civil & Environmental Engineering University of Alberta

More information

Using Blast Data to infer Training Images for MPS Simulation of Continuous Variables

Using Blast Data to infer Training Images for MPS Simulation of Continuous Variables Paper 34, CCG Annual Report 14, 212 ( 212) Using Blast Data to infer Training Images for MPS Simulation of Continuous Variables Hai T. Nguyen and Jeff B. Boisvert Multiple-point simulation (MPS) methods

More information

On internal consistency, conditioning and models of uncertainty

On internal consistency, conditioning and models of uncertainty On internal consistency, conditioning and models of uncertainty Jef Caers, Stanford University Abstract Recent research has been tending towards building models of uncertainty of the Earth, not just building

More information

Rotation and affinity invariance in multiple-point geostatistics

Rotation and affinity invariance in multiple-point geostatistics Rotation and ainity invariance in multiple-point geostatistics Tuanfeng Zhang December, 2001 Abstract Multiple-point stochastic simulation of facies spatial distribution requires a training image depicting

More information

HIERARCHICAL SIMULATION OF MULTIPLE-FACIES RESERVOIRS USING MULTIPLE-POINT GEOSTATISTICS

HIERARCHICAL SIMULATION OF MULTIPLE-FACIES RESERVOIRS USING MULTIPLE-POINT GEOSTATISTICS HIERARCHICAL SIMULATION OF MULTIPLE-FACIES RESERVOIRS USING MULTIPLE-POINT GEOSTATISTICS AREPORT SUBMITTED TO THE DEPARTMENT OF PETROLEUM ENGINEERING OF STANFORD UNIVERSITY IN PARTIAL FULFILLMENT OF THE

More information

Tensor Based Approaches for LVA Field Inference

Tensor Based Approaches for LVA Field Inference Tensor Based Approaches for LVA Field Inference Maksuda Lillah and Jeff Boisvert The importance of locally varying anisotropy (LVA) in model construction can be significant; however, it is often ignored

More information

A 3D code for mp simulation of continuous and

A 3D code for mp simulation of continuous and A 3D code for mp simulation of continuous and categorical variables: FILTERSIM Jianbing Wu, Alexandre Boucher & André G. Journel May, 2006 Abstract In most petroleum and geoscience studies, the flow is

More information

Direct Pattern-based Simulation of Non-Stationary Geostatistical Models

Direct Pattern-based Simulation of Non-Stationary Geostatistical Models Direct Pattern-based Simulation of Non-Stationary Geostatistical Models Mehrdad Honarkhah and Jef Caers Department of Energy Resources Engineering Stanford University Abstract Non-stationary models often

More information

Application of MPS Simulation with Multiple Training Image (MultiTI-MPS) to the Red Dog Deposit

Application of MPS Simulation with Multiple Training Image (MultiTI-MPS) to the Red Dog Deposit Application of MPS Simulation with Multiple Training Image (MultiTI-MPS) to the Red Dog Deposit Daniel A. Silva and Clayton V. Deutsch A Multiple Point Statistics simulation based on the mixing of two

More information

A workflow to account for uncertainty in well-log data in 3D geostatistical reservoir modeling

A workflow to account for uncertainty in well-log data in 3D geostatistical reservoir modeling A workflow to account for uncertainty in well-log data in 3D geostatistical reservoir Jose Akamine and Jef Caers May, 2007 Stanford Center for Reservoir Forecasting Abstract Traditionally well log data

More information

A009 HISTORY MATCHING WITH THE PROBABILITY PERTURBATION METHOD APPLICATION TO A NORTH SEA RESERVOIR

A009 HISTORY MATCHING WITH THE PROBABILITY PERTURBATION METHOD APPLICATION TO A NORTH SEA RESERVOIR 1 A009 HISTORY MATCHING WITH THE PROBABILITY PERTURBATION METHOD APPLICATION TO A NORTH SEA RESERVOIR B. Todd HOFFMAN and Jef CAERS Stanford University, Petroleum Engineering, Stanford CA 94305-2220 USA

More information

Modeling Multiple Rock Types with Distance Functions: Methodology and Software

Modeling Multiple Rock Types with Distance Functions: Methodology and Software Modeling Multiple Rock Types with Distance Functions: Methodology and Software Daniel A. Silva and Clayton V. Deutsch The sub division of the deposit into estimation domains that are internally consistent

More information

A Geostatistical and Flow Simulation Study on a Real Training Image

A Geostatistical and Flow Simulation Study on a Real Training Image A Geostatistical and Flow Simulation Study on a Real Training Image Weishan Ren (wren@ualberta.ca) Department of Civil & Environmental Engineering, University of Alberta Abstract A 12 cm by 18 cm slab

More information

Geostatistical modelling of offshore diamond deposits

Geostatistical modelling of offshore diamond deposits Geostatistical modelling of offshore diamond deposits JEF CAERS AND LUC ROMBOUTS STANFORD UNIVERSITY, Department of Petroleum Engineering, Stanford, CA 94305-2220, USA jef@pangea.stanford.edu TERRACONSULT,

More information

MPS Simulation with a Gibbs Sampler Algorithm

MPS Simulation with a Gibbs Sampler Algorithm MPS Simulation with a Gibbs Sampler Algorithm Steve Lyster and Clayton V. Deutsch Complex geologic structure cannot be captured and reproduced by variogram-based geostatistical methods. Multiple-point

More information

SIMPAT: Stochastic Simulation with Patterns

SIMPAT: Stochastic Simulation with Patterns SIMPAT: Stochastic Simulation with Patterns G. Burc Arpat Stanford Center for Reservoir Forecasting Stanford University, Stanford, CA 94305-2220 April 26, 2004 Abstract Flow in a reservoir is mostly controlled

More information

Iterative spatial resampling applied to seismic inverse modeling for lithofacies prediction

Iterative spatial resampling applied to seismic inverse modeling for lithofacies prediction Iterative spatial resampling applied to seismic inverse modeling for lithofacies prediction Cheolkyun Jeong, Tapan Mukerji, and Gregoire Mariethoz Department of Energy Resources Engineering Stanford University

More information

Improvements in Continuous Variable Simulation with Multiple Point Statistics

Improvements in Continuous Variable Simulation with Multiple Point Statistics Improvements in Continuous Variable Simulation with Multiple Point Statistics Jeff B. Boisvert A modified version of Mariethoz et al s (2010) algorithm for simulating continuous variables using multiple

More information

Geostatistics on Stratigraphic Grid

Geostatistics on Stratigraphic Grid Geostatistics on Stratigraphic Grid Antoine Bertoncello 1, Jef Caers 1, Pierre Biver 2 and Guillaume Caumon 3. 1 ERE department / Stanford University, Stanford CA USA; 2 Total CSTJF, Pau France; 3 CRPG-CNRS

More information

Hierarchical modeling of multi-scale flow barriers in channelized reservoirs

Hierarchical modeling of multi-scale flow barriers in channelized reservoirs Hierarchical modeling of multi-scale flow barriers in channelized reservoirs Hongmei Li and Jef Caers Stanford Center for Reservoir Forecasting Stanford University Abstract Channelized reservoirs often

More information

B. Todd Hoffman and Jef Caers Stanford University, California, USA

B. Todd Hoffman and Jef Caers Stanford University, California, USA Sequential Simulation under local non-linear constraints: Application to history matching B. Todd Hoffman and Jef Caers Stanford University, California, USA Introduction Sequential simulation has emerged

More information

Direct Sequential Co-simulation with Joint Probability Distributions

Direct Sequential Co-simulation with Joint Probability Distributions Math Geosci (2010) 42: 269 292 DOI 10.1007/s11004-010-9265-x Direct Sequential Co-simulation with Joint Probability Distributions Ana Horta Amílcar Soares Received: 13 May 2009 / Accepted: 3 January 2010

More information

History matching under training-image based geological model constraints

History matching under training-image based geological model constraints History matching under training-image based geological model constraints JEF CAERS Stanford University, Department of Petroleum Engineering Stanford, CA 94305-2220 January 2, 2002 Corresponding author

More information

Multiple-point geostatistics: a quantitative vehicle for integrating geologic analogs into multiple reservoir models

Multiple-point geostatistics: a quantitative vehicle for integrating geologic analogs into multiple reservoir models Multiple-point geostatistics: a quantitative vehicle for integrating geologic analogs into multiple reservoir models JEF CAERS AND TUANFENG ZHANG Stanford University, Stanford Center for Reservoir Forecasting

More information

Adaptive spatial resampling as a Markov chain Monte Carlo method for uncertainty quantification in seismic reservoir characterization

Adaptive spatial resampling as a Markov chain Monte Carlo method for uncertainty quantification in seismic reservoir characterization 1 Adaptive spatial resampling as a Markov chain Monte Carlo method for uncertainty quantification in seismic reservoir characterization Cheolkyun Jeong, Tapan Mukerji, and Gregoire Mariethoz Department

More information

Flexible Lag Definition for Experimental Variogram Calculation

Flexible Lag Definition for Experimental Variogram Calculation Flexible Lag Definition for Experimental Variogram Calculation Yupeng Li and Miguel Cuba The inference of the experimental variogram in geostatistics commonly relies on the method of moments approach.

More information

Programs for MDE Modeling and Conditional Distribution Calculation

Programs for MDE Modeling and Conditional Distribution Calculation Programs for MDE Modeling and Conditional Distribution Calculation Sahyun Hong and Clayton V. Deutsch Improved numerical reservoir models are constructed when all available diverse data sources are accounted

More information

High Resolution Geomodeling, Ranking and Flow Simulation at SAGD Pad Scale

High Resolution Geomodeling, Ranking and Flow Simulation at SAGD Pad Scale High Resolution Geomodeling, Ranking and Flow Simulation at SAGD Pad Scale Chad T. Neufeld, Clayton V. Deutsch, C. Palmgren and T. B. Boyle Increasing computer power and improved reservoir simulation software

More information

Selected Implementation Issues with Sequential Gaussian Simulation

Selected Implementation Issues with Sequential Gaussian Simulation Selected Implementation Issues with Sequential Gaussian Simulation Abstract Stefan Zanon (szanon@ualberta.ca) and Oy Leuangthong (oy@ualberta.ca) Department of Civil & Environmental Engineering University

More information

Conditioning a hybrid geostatistical model to wells and seismic data

Conditioning a hybrid geostatistical model to wells and seismic data Conditioning a hybrid geostatistical model to wells and seismic data Antoine Bertoncello, Gregoire Mariethoz, Tao Sun and Jef Caers ABSTRACT Hybrid geostatistical models imitate a sequence of depositional

More information

Markov Bayes Simulation for Structural Uncertainty Estimation

Markov Bayes Simulation for Structural Uncertainty Estimation P - 200 Markov Bayes Simulation for Structural Uncertainty Estimation Samik Sil*, Sanjay Srinivasan and Mrinal K Sen. University of Texas at Austin, samiksil@gmail.com Summary Reservoir models are built

More information

STOCHASTIC SIMULATION OF PATTERNS USING DISTANCE-BASED PATTERN MODELING

STOCHASTIC SIMULATION OF PATTERNS USING DISTANCE-BASED PATTERN MODELING STOCHASTIC SIMULATION OF PATTERNS USING DISTANCE-BASED PATTERN MODELING A DISSERTATION SUBMITTED TO THE DEPARTMENT OF ENERGY RESOURCES ENGINEERING AND THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY

More information

Exploring Direct Sampling and Iterative Spatial Resampling in History Matching

Exploring Direct Sampling and Iterative Spatial Resampling in History Matching Exploring Direct Sampling and Iterative Spatial Resampling in History Matching Matz Haugen, Grergoire Mariethoz and Tapan Mukerji Department of Energy Resources Engineering Stanford University Abstract

More information

Grid-less Simulation of a Fluvio-Deltaic Environment

Grid-less Simulation of a Fluvio-Deltaic Environment Grid-less Simulation of a Fluvio-Deltaic Environment R. Mohan Srivastava, FSS Canada Consultants, Toronto, Canada MoSrivastava@fssconsultants.ca and Marko Maucec, Halliburton Consulting and Project Management,

More information

SPE Copyright 2002, Society of Petroleum Engineers Inc.

SPE Copyright 2002, Society of Petroleum Engineers Inc. SPE 77958 Reservoir Modelling With Neural Networks And Geostatistics: A Case Study From The Lower Tertiary Of The Shengli Oilfield, East China L. Wang, S. Tyson, Geo Visual Systems Australia Pty Ltd, X.

More information

Sampling informative/complex a priori probability distributions using Gibbs sampling assisted by sequential simulation

Sampling informative/complex a priori probability distributions using Gibbs sampling assisted by sequential simulation Sampling informative/complex a priori probability distributions using Gibbs sampling assisted by sequential simulation Thomas Mejer Hansen, Klaus Mosegaard, and Knud Skou Cordua 1 1 Center for Energy Resources

More information

CONDITIONING FACIES SIMULATIONS WITH CONNECTIVITY DATA

CONDITIONING FACIES SIMULATIONS WITH CONNECTIVITY DATA CONDITIONING FACIES SIMULATIONS WITH CONNECTIVITY DATA PHILIPPE RENARD (1) and JEF CAERS (2) (1) Centre for Hydrogeology, University of Neuchâtel, Switzerland (2) Stanford Center for Reservoir Forecasting,

More information

Reservoir Modeling Combining Geostatistics with Markov Chain Monte Carlo Inversion

Reservoir Modeling Combining Geostatistics with Markov Chain Monte Carlo Inversion Reservoir Modeling Combining Geostatistics with Markov Chain Monte Carlo Inversion Andrea Zunino, Katrine Lange, Yulia Melnikova, Thomas Mejer Hansen and Klaus Mosegaard 1 Introduction Reservoir modeling

More information

Facies Modeling Using a Markov Mesh Model Specification

Facies Modeling Using a Markov Mesh Model Specification Math Geosci (2011) 43:611 624 DOI 10.1007/s11004-011-9350-9 Facies Modeling Using a Markov Mesh Model Specification Marita Stien Odd Kolbjørnsen Received: 28 April 2010 / Accepted: 13 February 2011 / Published

More information

Modeling Uncertainty in the Earth Sciences Jef Caers Stanford University

Modeling Uncertainty in the Earth Sciences Jef Caers Stanford University Modeling spatial continuity Modeling Uncertainty in the Earth Sciences Jef Caers Stanford University Motivation uncertain uncertain certain or uncertain uncertain Spatial Input parameters Spatial Stochastic

More information

The Proportional Effect: What it is and how do we model it?

The Proportional Effect: What it is and how do we model it? The Proportional Effect: What it is and how do we model it? John Manchuk Centre for Computational Geostatistics Department of Civil & Environmental Engineering University of Alberta One of the outstanding

More information

Geostatistical Reservoir Characterization of McMurray Formation by 2-D Modeling

Geostatistical Reservoir Characterization of McMurray Formation by 2-D Modeling Geostatistical Reservoir Characterization of McMurray Formation by 2-D Modeling Weishan Ren, Oy Leuangthong and Clayton V. Deutsch Department of Civil & Environmental Engineering, University of Alberta

More information

arxiv: v1 [eess.iv] 25 Dec 2018

arxiv: v1 [eess.iv] 25 Dec 2018 Vector Field-based Simulation of Tree-Like Non-Stationary Geostatistical Models arxiv:1812.11030v1 [eess.iv] 25 Dec 2018 Viviana Lorena Vargas, Sinesio Pesco Pontificia Universidade Catolica, Rio de Janeiro

More information

Final Exam Schedule. Final exam has been scheduled. 12:30 pm 3:00 pm, May 7. Location: INNOVA It will cover all the topics discussed in class

Final Exam Schedule. Final exam has been scheduled. 12:30 pm 3:00 pm, May 7. Location: INNOVA It will cover all the topics discussed in class Final Exam Schedule Final exam has been scheduled 12:30 pm 3:00 pm, May 7 Location: INNOVA 1400 It will cover all the topics discussed in class One page double-sided cheat sheet is allowed A calculator

More information

B002 DeliveryMassager - Propagating Seismic Inversion Information into Reservoir Flow Models

B002 DeliveryMassager - Propagating Seismic Inversion Information into Reservoir Flow Models B2 DeliveryMassager - Propagating Seismic Inversion Information into Reservoir Flow Models J. Gunning* (CSIRO Petroleum) & M.E. Glinsky (BHP Billiton Petroleum) SUMMARY We present a new open-source program

More information

A Geomodeling workflow used to model a complex carbonate reservoir with limited well control : modeling facies zones like fluid zones.

A Geomodeling workflow used to model a complex carbonate reservoir with limited well control : modeling facies zones like fluid zones. A Geomodeling workflow used to model a complex carbonate reservoir with limited well control : modeling facies zones like fluid zones. Thomas Jerome (RPS), Ke Lovan (WesternZagros) and Suzanne Gentile

More information

CONDITIONAL SIMULATION OF TRUNCATED RANDOM FIELDS USING GRADIENT METHODS

CONDITIONAL SIMULATION OF TRUNCATED RANDOM FIELDS USING GRADIENT METHODS CONDITIONAL SIMULATION OF TRUNCATED RANDOM FIELDS USING GRADIENT METHODS Introduction Ning Liu and Dean S. Oliver University of Oklahoma, Norman, Oklahoma, USA; ning@ou.edu The problem of estimating the

More information

Integration of Geostatistical Modeling with History Matching: Global and Regional Perturbation

Integration of Geostatistical Modeling with History Matching: Global and Regional Perturbation Integration of Geostatistical Modeling with History Matching: Global and Regional Perturbation Oliveira, Gonçalo Soares Soares, Amílcar Oliveira (CERENA/IST) Schiozer, Denis José (UNISIM/UNICAMP) Introduction

More information

The SPE Foundation through member donations and a contribution from Offshore Europe

The SPE Foundation through member donations and a contribution from Offshore Europe Primary funding is provided by The SPE Foundation through member donations and a contribution from Offshore Europe The Society is grateful to those companies that allow their professionals to serve as

More information

Pictures at an Exhibition

Pictures at an Exhibition Pictures at an Exhibition Han-I Su Department of Electrical Engineering Stanford University, CA, 94305 Abstract We employ an image identification algorithm for interactive museum guide with pictures taken

More information

Indicator Simulation Accounting for Multiple-Point Statistics

Indicator Simulation Accounting for Multiple-Point Statistics Indicator Simulation Accounting for Multiple-Point Statistics Julián M. Ortiz 1 and Clayton V. Deutsch 2 Geostatistical simulation aims at reproducing the variability of the real underlying phenomena.

More information

RM03 Integrating Petro-elastic Seismic Inversion and Static Model Building

RM03 Integrating Petro-elastic Seismic Inversion and Static Model Building RM03 Integrating Petro-elastic Seismic Inversion and Static Model Building P. Gelderblom* (Shell Global Solutions International BV) SUMMARY This presentation focuses on various aspects of how the results

More information

SCRF. 22 nd Annual Meeting. April 30-May

SCRF. 22 nd Annual Meeting. April 30-May SCRF 22 nd Annual Meeting April 30-May 1 2009 1 Research Overview CD annual report with papers Presentations 2 Modeling Uncertainty Distance based modeling of uncertainty - Model selection - Inverse modeling

More information

D025 Geostatistical Stochastic Elastic Iinversion - An Efficient Method for Integrating Seismic and Well Data Constraints

D025 Geostatistical Stochastic Elastic Iinversion - An Efficient Method for Integrating Seismic and Well Data Constraints D025 Geostatistical Stochastic Elastic Iinversion - An Efficient Method for Integrating Seismic and Well Data Constraints P.R. Williamson (Total E&P USA Inc.), A.J. Cherrett* (Total SA) & R. Bornard (CGGVeritas)

More information

CHAPTER 6 MODIFIED FUZZY TECHNIQUES BASED IMAGE SEGMENTATION

CHAPTER 6 MODIFIED FUZZY TECHNIQUES BASED IMAGE SEGMENTATION CHAPTER 6 MODIFIED FUZZY TECHNIQUES BASED IMAGE SEGMENTATION 6.1 INTRODUCTION Fuzzy logic based computational techniques are becoming increasingly important in the medical image analysis arena. The significant

More information

Schedule for Rest of Semester

Schedule for Rest of Semester Schedule for Rest of Semester Date Lecture Topic 11/20 24 Texture 11/27 25 Review of Statistics & Linear Algebra, Eigenvectors 11/29 26 Eigenvector expansions, Pattern Recognition 12/4 27 Cameras & calibration

More information

EE368 Project Report CD Cover Recognition Using Modified SIFT Algorithm

EE368 Project Report CD Cover Recognition Using Modified SIFT Algorithm EE368 Project Report CD Cover Recognition Using Modified SIFT Algorithm Group 1: Mina A. Makar Stanford University mamakar@stanford.edu Abstract In this report, we investigate the application of the Scale-Invariant

More information

Spatial Interpolation & Geostatistics

Spatial Interpolation & Geostatistics (Z i Z j ) 2 / 2 Spatial Interpolation & Geostatistics Lag Lag Mean Distance between pairs of points 1 Tobler s Law All places are related, but nearby places are related more than distant places Corollary:

More information

Efficient 3D Gravity and Magnetic Modeling

Efficient 3D Gravity and Magnetic Modeling Efficient 3D Gravity and Magnetic Modeling X. Li Fugro Gravity & Magnetic Services Inc., Houston, Texas, USA Summary There are many different spatial-domain algorithms for 3D gravity and magnetic forward

More information

A PARALLEL MODELLING APPROACH TO RESERVOIR CHARACTERIZATION

A PARALLEL MODELLING APPROACH TO RESERVOIR CHARACTERIZATION A PARALLEL MODELLING APPROACH TO RESERVOIR CHARACTERIZATION A DISSERTATION SUBMITTED TO THE DEPARTMENT OF PETROLEUM ENGINEERING AND THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY IN PARTIAL FULFILLMENT

More information

Parallel Multiple-Point Statistics Algorithm Based on List and Tree Structures

Parallel Multiple-Point Statistics Algorithm Based on List and Tree Structures Math Geosci (2013) 45:131 147 DOI 10.1007/s11004-012-9437-y Parallel Multiple-Point Statistics Algorithm Based on List and Tree Structures Julien Straubhaar Alexandre Walgenwitz Philippe Renard Received:

More information

Spatial Interpolation - Geostatistics 4/3/2018

Spatial Interpolation - Geostatistics 4/3/2018 Spatial Interpolation - Geostatistics 4/3/201 (Z i Z j ) 2 / 2 Spatial Interpolation & Geostatistics Lag Distance between pairs of points Lag Mean Tobler s Law All places are related, but nearby places

More information

An Accurate Method for Skew Determination in Document Images

An Accurate Method for Skew Determination in Document Images DICTA00: Digital Image Computing Techniques and Applications, 1 January 00, Melbourne, Australia. An Accurate Method for Skew Determination in Document Images S. Lowther, V. Chandran and S. Sridharan Research

More information

Variogram Inversion and Uncertainty Using Dynamic Data. Simultaneouos Inversion with Variogram Updating

Variogram Inversion and Uncertainty Using Dynamic Data. Simultaneouos Inversion with Variogram Updating Variogram Inversion and Uncertainty Using Dynamic Data Z. A. Reza (zreza@ualberta.ca) and C. V. Deutsch (cdeutsch@civil.ualberta.ca) Department of Civil & Environmental Engineering, University of Alberta

More information

Integrating 2-D, 3-D Yields New Insights

Integrating 2-D, 3-D Yields New Insights JULY 2007 The Better Business Publication Serving the Exploration / Drilling / Production Industry Integrating 2-D, 3-D Yields New Insights By Tony Rebec and Tony Marsh automatic fault tracking on sections,

More information

Using 3D-DEGA. Omer Inanc Tureyen and Jef Caers Department of Petroleum Engineering Stanford University

Using 3D-DEGA. Omer Inanc Tureyen and Jef Caers Department of Petroleum Engineering Stanford University Using 3D-DEGA Omer Inanc Tureyen and Jef Caers Department of Petroleum Engineering Stanford University 1 1 Introduction With the advance of CPU power, numerical reservoir models have become an essential

More information

Image Inpainting by Hyperbolic Selection of Pixels for Two Dimensional Bicubic Interpolations

Image Inpainting by Hyperbolic Selection of Pixels for Two Dimensional Bicubic Interpolations Image Inpainting by Hyperbolic Selection of Pixels for Two Dimensional Bicubic Interpolations Mehran Motmaen motmaen73@gmail.com Majid Mohrekesh mmohrekesh@yahoo.com Mojtaba Akbari mojtaba.akbari@ec.iut.ac.ir

More information

The Effect of Changing Grid Size in the Creation of Laser Scanner Digital Surface Models

The Effect of Changing Grid Size in the Creation of Laser Scanner Digital Surface Models The Effect of Changing Grid Size in the Creation of Laser Scanner Digital Surface Models Smith, S.L 1, Holland, D.A 1, and Longley, P.A 2 1 Research & Innovation, Ordnance Survey, Romsey Road, Southampton,

More information

We 2MIN 02 Data Density and Resolution Power in 3D DC Resistivity Surveys

We 2MIN 02 Data Density and Resolution Power in 3D DC Resistivity Surveys We 2MIN 02 Data Density and Resolution Power in 3D DC Resistivity Surveys M. Gharibi 1 *, R. Sharpe 1 1 Quantec Geoscience Ltd Summary Resolution power of a field 3D DC dataset is examined by gradually

More information

Th Volume Based Modeling - Automated Construction of Complex Structural Models

Th Volume Based Modeling - Automated Construction of Complex Structural Models Th-04-06 Volume Based Modeling - Automated Construction of Complex Structural Models L. Souche* (Schlumberger), F. Lepage (Schlumberger) & G. Iskenova (Schlumberger) SUMMARY A new technology for creating,

More information

CONDITIONING SURFACE-BASED MODELS TO WELL AND THICKNESS DATA

CONDITIONING SURFACE-BASED MODELS TO WELL AND THICKNESS DATA CONDITIONING SURFACE-BASED MODELS TO WELL AND THICKNESS DATA A DISSERTATION SUBMITTED TO THE DEPARTMENT OF ENERGY RESOURCES ENGINEERING AND THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY IN PARTIAL

More information

convolution shift invariant linear system Fourier Transform Aliasing and sampling scale representation edge detection corner detection

convolution shift invariant linear system Fourier Transform Aliasing and sampling scale representation edge detection corner detection COS 429: COMPUTER VISON Linear Filters and Edge Detection convolution shift invariant linear system Fourier Transform Aliasing and sampling scale representation edge detection corner detection Reading:

More information

ELEC Dr Reji Mathew Electrical Engineering UNSW

ELEC Dr Reji Mathew Electrical Engineering UNSW ELEC 4622 Dr Reji Mathew Electrical Engineering UNSW Review of Motion Modelling and Estimation Introduction to Motion Modelling & Estimation Forward Motion Backward Motion Block Motion Estimation Motion

More information

7. Vertical Layering

7. Vertical Layering 7.1 Make Horizons 7. Vertical Layering The vertical layering process consists of 4 steps: 1. Make Horizons: Insert the input surfaces into the 3D Grid. The inputs can be surfaces from seismic or well tops,

More information

surface but these local maxima may not be optimal to the objective function. In this paper, we propose a combination of heuristic methods: first, addi

surface but these local maxima may not be optimal to the objective function. In this paper, we propose a combination of heuristic methods: first, addi MetaHeuristics for a Non-Linear Spatial Sampling Problem Eric M. Delmelle Department of Geography and Earth Sciences University of North Carolina at Charlotte eric.delmelle@uncc.edu 1 Introduction In spatial

More information

Experiments with Edge Detection using One-dimensional Surface Fitting

Experiments with Edge Detection using One-dimensional Surface Fitting Experiments with Edge Detection using One-dimensional Surface Fitting Gabor Terei, Jorge Luis Nunes e Silva Brito The Ohio State University, Department of Geodetic Science and Surveying 1958 Neil Avenue,

More information

We G Application of Image-guided Interpolation to Build Low Frequency Background Model Prior to Inversion

We G Application of Image-guided Interpolation to Build Low Frequency Background Model Prior to Inversion We G106 05 Application of Image-guided Interpolation to Build Low Frequency Background Model Prior to Inversion J. Douma (Colorado School of Mines) & E. Zabihi Naeini* (Ikon Science) SUMMARY Accurate frequency

More information

Non Stationary Variograms Based on Continuously Varying Weights

Non Stationary Variograms Based on Continuously Varying Weights Non Stationary Variograms Based on Continuously Varying Weights David F. Machuca-Mory and Clayton V. Deutsch Centre for Computational Geostatistics Department of Civil & Environmental Engineering University

More information

Appropriate algorithm method for Petrophysical properties to construct 3D modeling for Mishrif formation in Amara oil field Jawad K.

Appropriate algorithm method for Petrophysical properties to construct 3D modeling for Mishrif formation in Amara oil field Jawad K. Appropriate algorithm method for Petrophysical properties to construct 3D modeling for Mishrif formation in Amara oil field Jawad K. Radhy AlBahadily Department of geology, college of science, Baghdad

More information

Texture Image Segmentation using FCM

Texture Image Segmentation using FCM Proceedings of 2012 4th International Conference on Machine Learning and Computing IPCSIT vol. 25 (2012) (2012) IACSIT Press, Singapore Texture Image Segmentation using FCM Kanchan S. Deshmukh + M.G.M

More information

Locating ego-centers in depth for hippocampal place cells

Locating ego-centers in depth for hippocampal place cells 204 5th Joint Symposium on Neural Computation Proceedings UCSD (1998) Locating ego-centers in depth for hippocampal place cells Kechen Zhang,' Terrence J. Sejeowski112 & Bruce L. ~cnau~hton~ 'Howard Hughes

More information

STOCHASTIC OBJECT-BASED SIMULATION OF CHANNELS CONSTRAINED ABSTRACT INTRODUCTION 1 DATA AND GEOLOGICAL FRAMEWORK BY HIGH RESOLUTION SEISMIC DATA

STOCHASTIC OBJECT-BASED SIMULATION OF CHANNELS CONSTRAINED ABSTRACT INTRODUCTION 1 DATA AND GEOLOGICAL FRAMEWORK BY HIGH RESOLUTION SEISMIC DATA STOCHASTIC OBJECT-BASED SIMULATION OF CHANNELS CONSTRAINED BY HIGH RESOLUTION SEISMIC DATA S. Viseur, A. Shtuka, J-L. Mallet ABSTRACT Recent progresses in exploration techniques (sonar images, high resolution

More information

Enhanced Iris Recognition System an Integrated Approach to Person Identification

Enhanced Iris Recognition System an Integrated Approach to Person Identification Enhanced Iris Recognition an Integrated Approach to Person Identification Gaganpreet Kaur Research Scholar, GNDEC, Ludhiana. Akshay Girdhar Associate Professor, GNDEC. Ludhiana. Manvjeet Kaur Lecturer,

More information

Cellular Learning Automata-Based Color Image Segmentation using Adaptive Chains

Cellular Learning Automata-Based Color Image Segmentation using Adaptive Chains Cellular Learning Automata-Based Color Image Segmentation using Adaptive Chains Ahmad Ali Abin, Mehran Fotouhi, Shohreh Kasaei, Senior Member, IEEE Sharif University of Technology, Tehran, Iran abin@ce.sharif.edu,

More information

DETECTION AND ROBUST ESTIMATION OF CYLINDER FEATURES IN POINT CLOUDS INTRODUCTION

DETECTION AND ROBUST ESTIMATION OF CYLINDER FEATURES IN POINT CLOUDS INTRODUCTION DETECTION AND ROBUST ESTIMATION OF CYLINDER FEATURES IN POINT CLOUDS Yun-Ting Su James Bethel Geomatics Engineering School of Civil Engineering Purdue University 550 Stadium Mall Drive, West Lafayette,

More information

EE795: Computer Vision and Intelligent Systems

EE795: Computer Vision and Intelligent Systems EE795: Computer Vision and Intelligent Systems Spring 2012 TTh 17:30-18:45 FDH 204 Lecture 14 130307 http://www.ee.unlv.edu/~b1morris/ecg795/ 2 Outline Review Stereo Dense Motion Estimation Translational

More information

Short Note. Non-stationary PEFs and large gaps. William Curry 1 INTRODUCTION

Short Note. Non-stationary PEFs and large gaps. William Curry 1 INTRODUCTION Stanford Exploration Project, Report 120, May 3, 2005, pages 247 257 Short Note Non-stationary PEFs and large gaps William Curry 1 INTRODUCTION Prediction-error filters (PEFs) may be used to interpolate

More information

Spatial Analysis and Modeling (GIST 4302/5302) Guofeng Cao Department of Geosciences Texas Tech University

Spatial Analysis and Modeling (GIST 4302/5302) Guofeng Cao Department of Geosciences Texas Tech University Spatial Analysis and Modeling (GIST 4302/5302) Guofeng Cao Department of Geosciences Texas Tech University 1 Outline of This Week Last topic, we learned: Spatial autocorrelation of areal data Spatial regression

More information

Templates, Image Pyramids, and Filter Banks

Templates, Image Pyramids, and Filter Banks Templates, Image Pyramids, and Filter Banks Computer Vision James Hays, Brown Slides: Hoiem and others Reminder Project due Friday Fourier Bases Teases away fast vs. slow changes in the image. This change

More information

Developing a Smart Proxy for the SACROC Water-Flooding Numerical Reservoir Simulation Model

Developing a Smart Proxy for the SACROC Water-Flooding Numerical Reservoir Simulation Model SPE-185691-MS Developing a Smart Proxy for the SACROC Water-Flooding Numerical Reservoir Simulation Model Faisal Alenezi and Shahab Mohaghegh, West Virginia University Copyright 2017, Society of Petroleum

More information

QUANTIZER DESIGN FOR EXPLOITING COMMON INFORMATION IN LAYERED CODING. Mehdi Salehifar, Tejaswi Nanjundaswamy, and Kenneth Rose

QUANTIZER DESIGN FOR EXPLOITING COMMON INFORMATION IN LAYERED CODING. Mehdi Salehifar, Tejaswi Nanjundaswamy, and Kenneth Rose QUANTIZER DESIGN FOR EXPLOITING COMMON INFORMATION IN LAYERED CODING Mehdi Salehifar, Tejaswi Nanjundaswamy, and Kenneth Rose Department of Electrical and Computer Engineering University of California,

More information

Chuanrong Zhang 1 and Weidong Li Department of Geography, Kent State University, Kent, Ohio 44242

Chuanrong Zhang 1 and Weidong Li Department of Geography, Kent State University, Kent, Ohio 44242 Comparing a Fixed-Path Markov Chain Geostatistical Algorithm with Sequential Indicator Simulation in Categorical Variable Simulation from Regular Samples Chuanrong Zhang 1 and Weidong Li Department of

More information

A low rank based seismic data interpolation via frequencypatches transform and low rank space projection

A low rank based seismic data interpolation via frequencypatches transform and low rank space projection A low rank based seismic data interpolation via frequencypatches transform and low rank space projection Zhengsheng Yao, Mike Galbraith and Randy Kolesar Schlumberger Summary We propose a new algorithm

More information

Using Similarity Attribute as a Quality Control Tool in 5D Interpolation

Using Similarity Attribute as a Quality Control Tool in 5D Interpolation Using Similarity Attribute as a Quality Control Tool in 5D Interpolation Muyi Kola-Ojo Launch Out Geophysical Services, Calgary, Alberta, Canada Summary Seismic attributes in the last two decades have

More information

MULTI ORIENTATION PERFORMANCE OF FEATURE EXTRACTION FOR HUMAN HEAD RECOGNITION

MULTI ORIENTATION PERFORMANCE OF FEATURE EXTRACTION FOR HUMAN HEAD RECOGNITION MULTI ORIENTATION PERFORMANCE OF FEATURE EXTRACTION FOR HUMAN HEAD RECOGNITION Panca Mudjirahardjo, Rahmadwati, Nanang Sulistiyanto and R. Arief Setyawan Department of Electrical Engineering, Faculty of

More information

2D Geostatistical Modeling and Volume Estimation of an Important Part of Western Onland Oil Field, India.

2D Geostatistical Modeling and Volume Estimation of an Important Part of Western Onland Oil Field, India. and Volume Estimation of an Important Part of Western Onland Oil Field, India Summary Satyajit Mondal*, Liendon Ziete, and B.S.Bisht ( GEOPIC, ONGC) M.A.Z.Mallik (E&D, Directorate, ONGC) Email: mondal_satyajit@ongc.co.in

More information

Multiple Model Estimation : The EM Algorithm & Applications

Multiple Model Estimation : The EM Algorithm & Applications Multiple Model Estimation : The EM Algorithm & Applications Princeton University COS 429 Lecture Dec. 4, 2008 Harpreet S. Sawhney hsawhney@sarnoff.com Plan IBR / Rendering applications of motion / pose

More information