Bradley C. Wallet

Latent space modeling of seismic data: An overview

Modeling of seismic data takes two forms: those based on physical or geological (phenomenological) models and those that are data-driven (probabilistic) models. In the phenomenological approach, physical and geologic models are tied to seismic data either through geologic analogs or principles of structural deformation and sedimentary deposition. The results are then compared to the observed data, and the model is iterated as necessary to improve agreement. In contrast, probabilistic modeling looks at patterns in the data. The data could include raw seismic observations or seismic attributes. Probabilities can then be assigned to observations or potential observations; however, many common techniques such as neural networks and clustering do not explicitly take this step.

Feasibility of using template-based and object-based automated detection methods for quantifying black and hybrid imported fire ant (Solenopsis invicta and S. invicta × richteri) mounds in aerial digital imagery

Imported fire ants construct earthen nests (mounds) that exhibit many characteristics which make them potentially good targets for remote sensing programs, including geographical orientation, topography, and bare soil surrounded by actively growing vegetation. Template-based features and object-based features extracted from aerial multispectral imagery of fire ant infested pastures were used to construct classifiers for automated fire ant mound detection. Aclassifierconstructedusingtemplate-basedfeaturesaloneyieldeda79%probabilityofdetectionwithacorrespondingfalse positiverateof9%.Additionofobject-basedfeatures(compactnessandsymmetry)totheclassifieryieldeda79%probability ofdetectionwithacorrespondingfalsepositiverateof4%.Maintaininga79%detectionratewhenapplyingtheclassifiertoa second, unique pasture dataset with different seasonal and other environmental factors resulted in a false positive rate of 17.5%. Data demonstrate that automated detection of mounds with classifiers incorporating template- and object-based features is feasible, but it may be necessary to construct unique classifiers on a site-specific basis.

Dynamic Thermal Structure of Imported Fire Ant Mounds

Abstract A study was undertaken to characterize surface temperatures of mounds of imported fire ant, Solenopsis invicta Buren (Hymenoptera: Formicidae) and S. richteri Forel, and their hybrid, as it relates to sun position and shape of the mounds, to better understand factors that affect absorption of solar radiation by the nest mound and to test feasibility of using thermal infrared imagery to remotely sense mounds. Mean mound surface temperature peaked shortly after solar noon and exceeded mean surface temperature of the surrounding surface. Temperature range for mounds and their surroundings peaked near solar noon, and the temperature range of the mound surface exceeded that of the surrounding area. The temperature difference between mounds and their surroundings peaked around solar noon and ranged from about 2 to 10°C. Quadratic trends relating temperature measurements to time of day (expressed as percentage of daylight hours from apparent sunrise to apparent sunset) explained 77 to 88% of the variation in the data. Mounds were asymmetrical, with the apex offset on average 81.5 ± 1.2 mm to the north of the average center. South facing aspects were about 20% larger than north facing aspects. Mound surface aspect and slope affected surface temperature; this affect was greatly influenced by time of day. Thermal infrared imagery was used to illustrate the effect of mound shape on surface temperature. These results indicate that the temperature differences between mounds and their surroundings are sufficient for detection using thermal infrared remote sensing, and predictable temporal changes in surface temperature may be useful for classifying mounds in images.

A grand tour of multispectral components: A tutorial

The human interpreter is perhaps the best pattern-recognition engine currently available. Often, interpreting seismic data involves inspection and analysis of multiple seismic attributes. While interpreters are skilled at visualizing complex geologic features in 3D, they are less adept at visualizing a great many attributes at once. Recent software development provides interpreters with the means of corendering three attributes (using a red-green-blue color model) or four attributes (using a hue-lightness-saturation-opacity color model). While analysis of four or more attributes is simple from a mathematical point of view, it is overly taxing on the normal human interpreter who needs to relate these images to the underlying geologic processes.

Attribute expression of channel forms in a hybrid carbonate turbidite formation

ABSTRACTMuch of the world’s conventional oil and gas production comes from turbidite systems. Interpreting them in three dimensions using commercially available software generally requires seismic attributes. Hybrid carbonate turbidite systems are an interesting phenomenon that is not fully understood. I have examined the attribute expression of channel forms in a hybrid carbonate turbidite system from off the coast of Western Australia. I have determined several characteristic responses to attributes that improve the ability to identify and delineate the channel forms. Finally, I have evaluated and developed a workflow that is effective at modeling and extracting the channel forms in three dimensions, leading to a product that can be used in further understanding of how carbonate turbidite systems develop.

Using the image grand tour to visualize fluvial deltaic architectural elements in south Texas, USA

AbstractSpectral decomposition can produce dozens of attributes for a single data set, far exceeding the ability for direct visualization. Some solutions have been proposed. The state-of-the-art approach is via the use of principal component analysis. However, this approach has significant inherent weaknesses, such as a lack of inclusion of spatial information and a tendency to inflate noise. Previous work has shown the ability of the image grand tour to construct lower-dimensional views of spectral information resulting in multiple images showing distinct architectural components. I propose a novel workflow for constructing color images to display multiple structures simultaneously. These images are constructed in a way that makes them complementary, leading to rich color images that are useful for interpretation. I demonstrate the value of this workflow though application to a land survey over Tertiary channels from south Texas.

Introduction to special section: Thin beds

Many thin beds in the subsurface are associated with hydrocarbons or other minerals. The seismic response of thin beds is complex due to interference between reflections from the different thin-layered interfaces. The conventional method of determining lithology, thickness, and other properties may

Integrating phase into the visualization of spectral decomposition attributes

AbstractMuch of the world’s conventional oil and gas production comes from fluvial-deltaic reservoirs. The ability to accurately interpret the architectural elements comprising these systems greatly reduces the risk in exploration and development in these environments. We have evaluated methods for using spectral decomposition attributes to improve the visualization in fluvial-deltaic environments using data from the Middle Pennsylvanian age Red Fork Formation of Oklahoma. We determined how spectral phase and magnitude attributes can be effectively combined using an hue-saturation-value color map to produce images that have considerable interpretational value. Incorporating our methods in the interpretation process has the potential to improve the exploration and development in these systems.

Using Diffusion Maps for Latent Space Analysis on Seismic Waveforms of Incised Valleys in the Anadarko Basin, Oklahoma, USA

In this work we present a new method for latent space mapping based upon inter-point similarities. This method, diffusion maps, has a number of nice qualities compared with previous methods including the fact that it is based upon inter-point similarities rather than a Euclidean space. We then demonstrate application of this approach to mapping an incised valley system from the Anadarko Basin, Oklahoma, USA. Multi-dimensional data is commonly encountered in attribute analysis where the desire is to combine several attributes with complementary properties. As geophysicists, we tend to view attribute spaces higher than four dimensional as being undesirable as they cannot be visualized using common color models such as ARGB space. Furthermore, mathematical considerations such as the Curse of Dimensionality (Bellmann, 1957) make working in lower dimensional space necessary. (Guo, Marfurt, Liu, & Dou, 2006) discussed an unsupervised learning method for doing dimensionality reduction of attributes using Principal Component Analysis (PCA). While this method has been shown to be useful in a broad range of applications, they are limited in their ability to capture non-linear structure in multidimensional attribute space. Retaining this non-linearity is important as data sets generally present a heterogeneous mix of latent processes that, taken as a whole, are unlikely to be well represented by a single lowdimensional linear manifold. A number of non-linear methods of manifold learning have been applied to seismic attributes and waveform modeling (Wallet & Marfurt, 2008) . Self organizing maps (SOM) is the best known of these approaches, and it is available in a number of commercial products. (Wallet & Perez, 2009) also demonstrated the use of a statistical method, generative topographical maps (GTM). (Wallet & Perez, 2009) applied diffusion maps to the problem of modeling well log data. They noted that the high computational demands of diffusion maps was an impediment to scaling to reasonable sized seismic problems. In this paper, we present an estimation method that deals with this problem.

Clustering Bed Sets from the Barnett Shale using Diffusion Map Attributes

Considerable effort has been invested in developing methods for classifying and identifying facies from well logs. However, identifying facies at individual measurement points tells us little about the processes that gave rise to these individual facies. In this paper, we propose a method for clustering and analyzing bed sets from well logs. The main innovation is the use of a nonlinear method for manifold learning based upon interpoint distances. We then apply this method to gamma ray logs from the Barnett Shale and show the results. Finally, we discuss how this method may be used in workflows for hydrocarbon exploration and development.