Enhancing Seismic Data Resolution With Multi-Attribute Analysis Using Both Well Log Data and Seismic Data – A Case Study

Abstract

The available well locations for a 3D survey area are often irregularly spaced and may even be few in number. 3D seismic data has to be relied on to generate a 3D volume of various log properties such as P-wave velocity, density, or porosity between well locations. Well log curves have very fine sampling in the vertical direction and hence good resolution, but lithologic properties are desired spatially for reservoir characterization and prospect generation. Seismic data possess lower resolution but good spatial sampling and so can be used for the purpose. Therefore, the ability to improve the seismic resolution through multi-attribute analysis followed by neural network-type processes and thereby improve the details of the derived log property is a definite advantage in terms of determining new drilling locations in a 3D survey area. The target logs are porosity logs which are commonly available at the well locations and these logs can be used to correlate with various seismic attributes at each well location on a sample-by-sample basis. The correlation process derives relationships between the target log data, in this case porosity logs and seismic attributes and is referred to as the “training” stage. A subset of the available target logs are used in the training process and the relationships that are derived during the training stage of the process are used in the “application” stage of the process to predict the log property of porosity throughout the 3D survey area. The Blackfoot field, is a glauconitic compound-incised valley system comprised of three cycles of incision with the upper and lower incised valleys being the main reservoirs. 12 wells within the area of 3D seismic data will be used in this project. A comparison of the porosity log curves from the multi-attribute transform and the Probabilistic Neural Network (PNN) with the actual log data will illustrates the ability of the PNN workflow to enhance the seismic resolution in terms of the lithologic property i.e. porosity. Various displays such as contour maps, data slices in time, as well as along horizons, coherence, and colour enhanced decomposition slices will be used to illustrate the channel fill system as derived from the multi-attribute analysis and probabilistic neural network application.

AAPG Datapages/Search and Discovery Article #90291 ©2017 AAPG Annual Convention and Exhibition, Houston, Texas, April 2-5, 2017